Fix safe user id digest algorithm

The default `make up` started the Gateway with `--workers 4`, but run state
(RunManager and the stream bridge) is held in-process and nginx uses no sticky
sessions. With the default config, same-run requests scatter across workers that
each keep their own run state, breaking run cancellation (409), SSE reconnect
(hangs on heartbeats), multitask de-duplication, and IM channels (duplicate
replies). The shared cross-worker stream bridge does not exist yet.

Default GATEWAY_WORKERS to 1 so the out-of-the-box deployment is correct,
document the single-worker boundary in the README, and add a regression test
pinning the default while keeping it overridable. This is a stop-gap, not a
multi-worker implementation; the full fix (shared run state + stream bridge) is
tracked in #3191.

Refs #3239, #3260

.env.example

@@ -21,6 +21,7 @@ INFOQUEST_API_KEY=your-infoquest-api-key
 # DEEPSEEK_API_KEY=your-deepseek-api-key
 # NOVITA_API_KEY=your-novita-api-key  # OpenAI-compatible, see https://novita.ai
 # MINIMAX_API_KEY=your-minimax-api-key  # OpenAI-compatible, see https://platform.minimax.io
+# STEPFUN_API_KEY=your-stepfun-api-key  # OpenAI-compatible, see https://platform.stepfun.com
 # VLLM_API_KEY=your-vllm-api-key  # OpenAI-compatible
 # FEISHU_APP_ID=your-feishu-app-id
 # FEISHU_APP_SECRET=your-feishu-app-secret

.github/labeler.yml

@@ -1,72 +0,0 @@
-# Path-based PR auto-labeling config for actions/labeler@v5.
-# Each key is a label (must exist — see .github/labels.yml); the globs decide
-# when it is applied. A PR can match several areas, which is expected.
-
-"area:frontend":
-  - changed-files:
-      - any-glob-to-any-file:
-          - "frontend/**"
-
-"area:backend":
-  - changed-files:
-      - any-glob-to-any-file:
-          - "backend/app/**"
-          - "backend/packages/harness/deerflow/runtime/**"
-          - "backend/packages/harness/deerflow/persistence/**"
-          - "backend/packages/harness/deerflow/config/**"
-          - "backend/packages/harness/deerflow/tools/**"
-          - "backend/packages/harness/deerflow/guardrails/**"
-          - "backend/packages/harness/deerflow/tracing/**"
-          - "backend/packages/harness/deerflow/models/**"
-          - "backend/packages/harness/deerflow/utils/**"
-          - "backend/packages/harness/deerflow/uploads/**"
-
-"area:agents":
-  - changed-files:
-      - any-glob-to-any-file:
-          - "backend/packages/harness/deerflow/agents/**"
-          - "backend/packages/harness/deerflow/subagents/**"
-          - "backend/packages/harness/deerflow/reflection/**"
-          - "backend/langgraph.json"
-          - "backend/**/prompts/**"
-
-"area:sandbox":
-  - changed-files:
-      - any-glob-to-any-file:
-          - "docker/**"
-          - "backend/packages/harness/deerflow/sandbox/**"
-          - "backend/Dockerfile"
-          - "frontend/Dockerfile"
-
-"area:skills":
-  - changed-files:
-      - any-glob-to-any-file:
-          - "skills/**"
-          - "backend/packages/harness/deerflow/skills/**"
-          - "frontend/src/core/skills/**"
-
-"area:mcp":
-  - changed-files:
-      - any-glob-to-any-file:
-          - "backend/packages/harness/deerflow/mcp/**"
-          - "frontend/src/core/mcp/**"
-
-"area:ci":
-  - changed-files:
-      - any-glob-to-any-file:
-          - ".github/**"
-          - "scripts/**"
-
-"area:docs":
-  - changed-files:
-      - any-glob-to-any-file:
-          - "docs/**"
-          - "**/*.md"
-
-"area:deps":
-  - changed-files:
-      - any-glob-to-any-file:
-          - "backend/pyproject.toml"
-          - "backend/uv.lock"
-          - "frontend/package.json"
-          - "frontend/pnpm-lock.yaml"

.github/workflows/issue-triage.yml

@@ -1,44 +0,0 @@
-name: Issue Triage
-
-# Ensures every newly opened issue carries `needs-triage`, even blank or
-# API-created ones that bypass the issue templates. Creates the label if it is
-# somehow missing, so the workflow is self-healing.
-
-on:
-  issues:
-    types: [opened]
-
-permissions:
-  issues: write
-
-jobs:
-  needs-triage:
-    runs-on: ubuntu-latest
-    steps:
-      - name: Add needs-triage label
-        uses: actions/github-script@v7
-        with:
-          script: |
-            const { owner, repo } = context.repo;
-            const issue_number = context.payload.issue.number;
-
-            const current = (context.payload.issue.labels || []).map(l => l.name);
-            if (current.includes('needs-triage')) {
-              core.info('Issue already has needs-triage; nothing to do.');
-              return;
-            }
-
-            // Self-heal: create the label if it does not exist yet.
-            try {
-              await github.rest.issues.createLabel({
-                owner, repo, name: 'needs-triage', color: 'fef2c0',
-                description: 'Awaiting maintainer triage',
-              });
-            } catch (e) {
-              if (e.status !== 422) throw e; // 422 = already exists
-            }
-
-            await github.rest.issues.addLabels({
-              owner, repo, issue_number, labels: ['needs-triage'],
-            });
-            core.info(`Added needs-triage to #${issue_number}.`);

.github/workflows/pr-labeler.yml

@@ -1,28 +0,0 @@
-name: PR Labeler
-
-# Applies area:* labels based on which files a PR changes (see .github/labeler.yml).
-# Uses pull_request_target so it also works on fork PRs. SAFE: actions/labeler
-# only reads the changed-file list via the API — it never checks out or runs PR code.
-
-on:
-  pull_request_target:
-    types: [opened, synchronize, reopened, ready_for_review]
-
-permissions:
-  contents: read
-  pull-requests: write
-
-concurrency:
-  group: pr-labeler-${{ github.event.pull_request.number }}
-  cancel-in-progress: true
-
-jobs:
-  label:
-    if: github.event.pull_request.draft == false
-    runs-on: ubuntu-latest
-    steps:
-      - name: Apply area labels
-        uses: actions/labeler@v5
-        with:
-          configuration-path: .github/labeler.yml
-          sync-labels: true

.github/workflows/pr-triage.yml

@@ -1,164 +0,0 @@
-name: PR Triage
-
-# Two responsibilities, both pure-metadata (no PR code is checked out or run):
-#   1. On open/sync: apply size/* + risk:* labels, and needs-validation when the
-#      PR touches the front/back contract surface (backend API, SSE, agents, or
-#      the frontend streaming client). A `skip-validation` label opts out.
-#   2. On maintainer review: apply the `reviewing` label.
-#
-# All labels are managed within their own namespace — labels outside size/*,
-# risk:*, needs-validation and reviewing are never touched here.
-
-on:
-  pull_request_target:
-    types: [opened, synchronize, reopened, ready_for_review]
-  pull_request_review:
-    types: [submitted]
-
-permissions:
-  contents: read
-  pull-requests: write
-
-concurrency:
-  group: pr-triage-${{ github.event.pull_request.number }}
-  cancel-in-progress: false
-
-jobs:
-  size-and-risk:
-    if: github.event_name == 'pull_request_target' && github.event.pull_request.draft == false
-    runs-on: ubuntu-latest
-    steps:
-      - name: Label size, risk and validation need
-        uses: actions/github-script@v7
-        with:
-          script: |
-            const pr = context.payload.pull_request;
-            const { owner, repo } = context.repo;
-            const prNumber = pr.number;
-
-            // ---- size, from additions + deletions ----
-            const churn = (pr.additions || 0) + (pr.deletions || 0);
-            const sizeLabel =
-              churn < 20 ? 'size/XS' :
-              churn < 100 ? 'size/S' :
-              churn < 300 ? 'size/M' :
-              churn < 700 ? 'size/L' : 'size/XL';
-
-            // ---- changed paths ----
-            const files = await github.paginate(github.rest.pulls.listFiles, {
-              owner, repo, pull_number: prNumber, per_page: 100,
-            });
-            const paths = files.map(f => f.filename);
-
-            const matches = (re) => paths.some(p => re.test(p));
-
-            const docsOnly = paths.length > 0 && paths.every(p =>
-              /\.(md|mdx|txt)$/i.test(p) || p.startsWith('docs/') ||
-              /\.(png|jpe?g|gif|svg|webp|ico)$/i.test(p));
-
-            const highRisk = matches(
-              /^backend\/app\/gateway\//) || matches(
-              /^backend\/packages\/harness\/deerflow\/(agents|subagents|sandbox)\//) || matches(
-              /(^|\/)langgraph\.json$/) || matches(
-              /(^|\/)(auth|authz|security)/i) || matches(
-              /(pyproject\.toml|uv\.lock|package\.json|pnpm-lock\.yaml)$/) || matches(
-              /^docker\//) || matches(
-              /^\.github\/workflows\//);
-
-            const riskLabel = docsOnly ? 'risk:low' : (highRisk ? 'risk:high' : 'risk:medium');
-
-            // needs-validation: front/back contract surface
-            const contractSurface =
-              matches(/^backend\/app\/gateway\//) ||
-              matches(/^backend\/packages\/harness\/deerflow\/(agents|subagents)\//) ||
-              matches(/(^|\/)langgraph\.json$/) ||
-              matches(/^frontend\/src\/core\/(api|threads|messages)\//);
-
-            const current = (pr.labels || []).map(l => l.name);
-            const hasSkip = current.includes('skip-validation');
-
-            const desired = [sizeLabel, riskLabel];
-            if (contractSurface && !hasSkip) desired.push('needs-validation');
-
-            const managed = (name) =>
-              name.startsWith('size/') || name.startsWith('risk:') || name === 'needs-validation';
-
-            const toRemove = current.filter(l => managed(l) && !desired.includes(l));
-            const toAdd = desired.filter(l => !current.includes(l));
-
-            for (const name of toRemove) {
-              try {
-                await github.rest.issues.removeLabel({ owner, repo, issue_number: prNumber, name });
-              } catch (e) {
-                if (e.status !== 404) throw e;
-              }
-            }
-            if (toAdd.length) {
-              await github.rest.issues.addLabels({ owner, repo, issue_number: prNumber, labels: toAdd });
-            }
-            core.info(`size=${sizeLabel} risk=${riskLabel} churn=${churn} ` +
-              `validation=${desired.includes('needs-validation')} ` +
-              `(+${toAdd.join(',') || '-'} / -${toRemove.join(',') || '-'})`);
-
-  first-time:
-    if: github.event_name == 'pull_request_target' && github.event.action == 'opened'
-    runs-on: ubuntu-latest
-    steps:
-      - name: Label first-time contributors
-        uses: actions/github-script@v7
-        with:
-          script: |
-            const pr = context.payload.pull_request;
-            const { owner, repo } = context.repo;
-            const assoc = pr.author_association;
-            const isBot = pr.user.type === 'Bot';
-            core.info(`author=${pr.user.login} association=${assoc} bot=${isBot}`);
-
-            // FIRST_TIME_CONTRIBUTOR = no prior merged commit to this repo;
-            // FIRST_TIMER = no prior commit anywhere on GitHub. Either counts.
-            if (isBot || !['FIRST_TIME_CONTRIBUTOR', 'FIRST_TIMER'].includes(assoc)) {
-              core.info('Not a first-time contributor; skipping.');
-              return;
-            }
-            await github.rest.issues.addLabels({
-              owner, repo, issue_number: pr.number, labels: ['first-time-contributor'],
-            });
-            core.info(`Added first-time-contributor to #${pr.number}.`);
-
-  reviewing:
-    if: github.event_name == 'pull_request_review'
-    runs-on: ubuntu-latest
-    steps:
-      - name: Add reviewing label for maintainer reviews
-        uses: actions/github-script@v7
-        with:
-          script: |
-            const { owner, repo } = context.repo;
-            const prNumber = context.payload.pull_request.number;
-            const reviewer = context.payload.review.user.login;
-
-            const { data: perm } = await github.rest.repos.getCollaboratorPermissionLevel({
-              owner, repo, username: reviewer,
-            });
-            if (!['admin', 'write', 'maintain'].includes(perm.permission)) {
-              core.info(`Reviewer ${reviewer} (${perm.permission}) is not a maintainer; skipping.`);
-              return;
-            }
-
-            const { data: labels } = await github.rest.issues.listLabelsOnIssue({
-              owner, repo, issue_number: prNumber,
-            });
-            if (labels.some(l => l.name === 'reviewing')) {
-              core.info('Already labeled reviewing; skipping.');
-              return;
-            }
-            try {
-              await github.rest.issues.addLabels({
-                owner, repo, issue_number: prNumber, labels: ['reviewing'],
-              });
-              core.info(`Added "reviewing" (reviewer ${reviewer}).`);
-            } catch (e) {
-              // 403 is expected for review events on some fork PR contexts.
-              if (e.status === 403) core.info('No permission to label (expected on some fork PRs).');
-              else throw e;
-            }

.github/workflows/replay-e2e.yml

@@ -0,0 +1,108 @@
+name: Replay E2E (front-back contract)
+
+# Guards the front-back contract via record/replay (no API key in CI):
+#   Layer 1 — backend golden: replay a recorded trace through the real gateway,
+#             assert the SSE event sequence matches the committed golden.
+#   Layer 2 — full-stack render: real Next.js frontend + real gateway (replay
+#             model) + Chromium; assert the replayed turns render in the browser.
+# Triggered by changes on EITHER side of the contract so a backend change can no
+# longer pass without the frontend-facing checks running.
+
+on:
+  push:
+    branches: ["main"]
+    paths:
+      - "frontend/**"
+      - "backend/app/gateway/**"
+      - "backend/packages/harness/**"
+      - "backend/tests/fixtures/replay/**"
+      - "backend/tests/replay_provider.py"
+      - "backend/tests/_replay_fixture.py"
+      - "backend/tests/seed_runs_router.py"
+      - "backend/tests/test_replay_golden.py"
+      - "backend/scripts/run_replay_gateway.py"
+      - ".github/workflows/replay-e2e.yml"
+  pull_request:
+    types: [opened, synchronize, reopened, ready_for_review]
+    paths:
+      - "frontend/**"
+      - "backend/app/gateway/**"
+      - "backend/packages/harness/**"
+      - "backend/tests/fixtures/replay/**"
+      - "backend/tests/replay_provider.py"
+      - "backend/tests/_replay_fixture.py"
+      - "backend/tests/seed_runs_router.py"
+      - "backend/tests/test_replay_golden.py"
+      - "backend/scripts/run_replay_gateway.py"
+      - ".github/workflows/replay-e2e.yml"
+
+concurrency:
+  group: replay-e2e-${{ github.event.pull_request.number || github.ref }}
+  cancel-in-progress: true
+
+permissions:
+  contents: read
+
+jobs:
+  backend-replay-golden:
+    name: Layer 1 — backend golden (no API key)
+    if: github.event_name != 'pull_request' || github.event.pull_request.draft == false
+    runs-on: ubuntu-latest
+    timeout-minutes: 15
+    steps:
+      - uses: actions/checkout@v6
+      - name: Set up Python
+        uses: actions/setup-python@v6
+        with:
+          python-version: "3.12"
+      - name: Install uv
+        uses: astral-sh/setup-uv@v7
+      - name: Install backend dependencies
+        working-directory: backend
+        run: uv sync --group dev
+      - name: Replay golden (backend SSE contract)
+        working-directory: backend
+        run: PYTHONPATH=. uv run pytest tests/test_replay_golden.py -v
+
+  fullstack-replay-render:
+    name: Layer 2 — full-stack render (no API key)
+    if: github.event_name != 'pull_request' || github.event.pull_request.draft == false
+    runs-on: ubuntu-latest
+    timeout-minutes: 25
+    steps:
+      - uses: actions/checkout@v6
+      - name: Set up Python
+        uses: actions/setup-python@v6
+        with:
+          python-version: "3.12"
+      - name: Install uv
+        uses: astral-sh/setup-uv@v7
+      - name: Install backend dependencies (replay gateway)
+        working-directory: backend
+        run: uv sync --group dev
+      - name: Setup Node.js
+        uses: actions/setup-node@v4
+        with:
+          node-version: "22"
+      - name: Enable Corepack
+        run: corepack enable
+      - name: Use pinned pnpm version
+        run: corepack prepare pnpm@10.26.2 --activate
+      - name: Install frontend dependencies
+        working-directory: frontend
+        run: pnpm install --frozen-lockfile
+      - name: Install Playwright Chromium
+        working-directory: frontend
+        run: npx playwright install chromium --with-deps
+      - name: Full-stack replay render (DOM assertions are the gate)
+        working-directory: frontend
+        run: pnpm exec playwright test -c playwright.real-backend.config.ts
+      - name: Upload report + render artifact
+        uses: actions/upload-artifact@v4
+        if: ${{ !cancelled() }}
+        with:
+          name: replay-render
+          path: |
+            frontend/playwright-report/
+            frontend/test-results/
+          retention-days: 7

.github/workflows/triage.yml

@@ -0,0 +1,223 @@
+name: Triage
+
+# One workflow for all event-driven PR/issue labeling. Replaces the former
+# pr-labeler / pr-triage / issue-triage workflows (and drops actions/labeler).
+#
+# Design notes:
+#   * All jobs are pure-metadata: they read changed-file lists / PR fields / the
+#     review payload via the API and write labels. PR code is NEVER checked out
+#     or executed, so pull_request_target is safe here.
+#   * Each job only reconciles labels in namespaces IT owns
+#     (area:* / size/* / risk:* / needs-validation). It never touches labels
+#     applied by maintainers or other tools (bug, priority, etc.). first-time-
+#     contributor and reviewing are add-only.
+#   * State is read LIVE (listFiles + listLabelsOnIssue) at run time, not from
+#     the (stale) event payload, so rapid synchronize events converge instead
+#     of thrashing.
+
+on:
+  pull_request_target:
+    types: [opened, synchronize, reopened, ready_for_review]
+  pull_request_review:
+    types: [submitted]
+  issues:
+    types: [opened]
+
+permissions:
+  contents: read
+  pull-requests: write
+  issues: write
+
+jobs:
+  # ── PR: area / size / risk / needs-validation / first-time ─────────────────
+  pr-labels:
+    if: github.event_name == 'pull_request_target' && github.event.pull_request.draft == false
+    runs-on: ubuntu-latest
+    concurrency:
+      group: triage-pr-${{ github.event.pull_request.number }}
+      cancel-in-progress: true
+    steps:
+      - name: Apply PR labels from live state
+        uses: actions/github-script@v8
+        with:
+          script: |
+            const pr = context.payload.pull_request;
+            const { owner, repo } = context.repo;
+            const num = pr.number;
+
+            // ---- live changed files ----
+            const files = await github.paginate(github.rest.pulls.listFiles, {
+              owner, repo, pull_number: num, per_page: 100,
+            });
+            const paths = files.map(f => f.filename);
+            const m = (re) => paths.some(p => re.test(p));
+
+            // ---- area: replaces .github/labeler.yml (path -> area) ----
+            const AREA_RULES = [
+              ['area:frontend', [/^frontend\//]],
+              ['area:backend',  [/^backend\/app\//, /^backend\/packages\/harness\/deerflow\/(runtime|persistence|config|tools|guardrails|tracing|models|utils|uploads)\//]],
+              ['area:agents',   [/^backend\/packages\/harness\/deerflow\/(agents|subagents|reflection)\//, /(^|\/)langgraph\.json$/, /^backend\/.*\/prompts\//]],
+              ['area:sandbox',  [/^docker\//, /^backend\/packages\/harness\/deerflow\/sandbox\//, /(^|\/)Dockerfile$/]],
+              ['area:skills',   [/^skills\//, /^backend\/packages\/harness\/deerflow\/skills\//, /^frontend\/src\/core\/skills\//]],
+              ['area:mcp',      [/^backend\/packages\/harness\/deerflow\/mcp\//, /^frontend\/src\/core\/mcp\//]],
+              ['area:ci',       [/^\.github\//, /^scripts\//]],
+              ['area:docs',     [/^docs\//, /\.mdx?$/]],
+              ['area:deps',     [/(^|\/)(pyproject\.toml|uv\.lock|package\.json|pnpm-lock\.yaml)$/]],
+            ];
+            const areaLabels = AREA_RULES
+              .filter(([, res]) => res.some(re => m(re)))
+              .map(([label]) => label);
+
+            // ---- size: additions+deletions, excluding lockfiles/snapshots ----
+            const EXCLUDE_SIZE = /(^|\/)(uv\.lock|pnpm-lock\.yaml|package-lock\.json)$|\.snap$/;
+            const churn = files
+              .filter(f => !EXCLUDE_SIZE.test(f.filename))
+              .reduce((s, f) => s + (f.additions || 0) + (f.deletions || 0), 0);
+            const sizeLabel =
+              churn < 20 ? 'size/XS' :
+              churn < 100 ? 'size/S' :
+              churn < 300 ? 'size/M' :
+              churn < 700 ? 'size/L' : 'size/XL';
+
+            // ---- risk ----
+            const docsOnly = paths.length > 0 && paths.every(p =>
+              /\.(md|mdx|txt)$/i.test(p) || p.startsWith('docs/') ||
+              /\.(png|jpe?g|gif|svg|webp|ico)$/i.test(p));
+            const highRisk =
+              m(/^backend\/app\/gateway\//) ||
+              m(/^backend\/packages\/harness\/deerflow\/(agents|subagents|sandbox)\//) ||
+              m(/(^|\/)langgraph\.json$/) ||
+              m(/(^|\/)(auth|authz|security)/i) ||
+              m(/(pyproject\.toml|uv\.lock|package\.json|pnpm-lock\.yaml)$/) ||
+              m(/^docker\//) ||
+              m(/^\.github\/workflows\//);
+            const riskLabel = docsOnly ? 'risk:low' : (highRisk ? 'risk:high' : 'risk:medium');
+
+            // ---- needs-validation: front/back contract surface ----
+            const contract =
+              m(/^backend\/app\/gateway\//) ||
+              m(/^backend\/packages\/harness\/deerflow\/(agents|subagents)\//) ||
+              m(/(^|\/)langgraph\.json$/) ||
+              m(/^frontend\/src\/core\/(api|threads|messages)\//);
+
+            // ---- live current labels (NOT the stale event payload) ----
+            const current = (await github.paginate(github.rest.issues.listLabelsOnIssue, {
+              owner, repo, issue_number: num, per_page: 100,
+            })).map(l => l.name);
+            const hasSkip = current.includes('skip-validation');
+
+            // Reconcile ONLY namespaces we own; never touch others.
+            const owned = (n) =>
+              n.startsWith('area:') || n.startsWith('size/') ||
+              n.startsWith('risk:') || n === 'needs-validation';
+            const desired = new Set([...areaLabels, sizeLabel, riskLabel]);
+            if (contract && !hasSkip) desired.add('needs-validation');
+
+            const toRemove = current.filter(n => owned(n) && !desired.has(n));
+            const toAdd = [...desired].filter(n => !current.includes(n));
+
+            // first-time-contributor: add-only, on opened, real users only.
+            if (context.payload.action === 'opened' &&
+                pr.user.type === 'User' &&
+                ['FIRST_TIME_CONTRIBUTOR', 'FIRST_TIMER'].includes(pr.author_association) &&
+                !current.includes('first-time-contributor')) {
+              toAdd.push('first-time-contributor');
+            }
+
+            for (const name of toRemove) {
+              try {
+                await github.rest.issues.removeLabel({ owner, repo, issue_number: num, name });
+              } catch (e) {
+                if (e.status !== 404) throw e;
+              }
+            }
+            if (toAdd.length) {
+              await github.rest.issues.addLabels({ owner, repo, issue_number: num, labels: toAdd });
+            }
+            core.info(`area=[${areaLabels.join(',')}] ${sizeLabel} ${riskLabel} churn=${churn} ` +
+              `validation=${desired.has('needs-validation')} ` +
+              `(+${toAdd.join(',') || '-'} / -${toRemove.join(',') || '-'})`);
+
+  # ── PR: reviewing label on a maintainer's human review ─────────────────────
+  reviewing:
+    if: github.event_name == 'pull_request_review'
+    runs-on: ubuntu-latest
+    concurrency:
+      group: triage-review-${{ github.event.pull_request.number }}
+      cancel-in-progress: false
+    steps:
+      - name: Add reviewing label for maintainer reviews
+        uses: actions/github-script@v8
+        with:
+          script: |
+            const { owner, repo } = context.repo;
+            const num = context.payload.pull_request.number;
+            const review = context.payload.review;
+            const assoc = review.author_association;     // payload field; no API call
+            const type = review.user && review.user.type;
+
+            // author_association is NONE for every automated reviewer
+            // (Copilot, CodeRabbit, Codex, Sourcery, ...), so this allowlist
+            // drops them all without a denylist — and never calls the
+            // collaborators API that 404s on "Copilot is not a user".
+            // user.type === 'User' guards the rare bot-added-as-collaborator case.
+            if (!['OWNER', 'MEMBER', 'COLLABORATOR'].includes(assoc) || type !== 'User') {
+              core.info(`reviewer ${review.user && review.user.login} assoc=${assoc} type=${type}; skipping.`);
+              return;
+            }
+
+            const labels = (await github.paginate(github.rest.issues.listLabelsOnIssue, {
+              owner, repo, issue_number: num, per_page: 100,
+            })).map(l => l.name);
+            if (labels.includes('reviewing')) {
+              core.info('Already labeled reviewing; skipping.');
+              return;
+            }
+            try {
+              await github.rest.issues.addLabels({
+                owner, repo, issue_number: num, labels: ['reviewing'],
+              });
+              core.info('Added "reviewing".');
+            } catch (e) {
+              if (e.status === 403) core.info('No permission to label (expected on some fork PRs).');
+              else throw e;
+            }
+
+  # ── Issue: needs-triage on every new issue ────────────────────────────────
+  issue-triage:
+    if: github.event_name == 'issues'
+    runs-on: ubuntu-latest
+    concurrency:
+      group: triage-issue-${{ github.event.issue.number }}
+      cancel-in-progress: false
+    steps:
+      - name: Add needs-triage label
+        uses: actions/github-script@v8
+        with:
+          script: |
+            const { owner, repo } = context.repo;
+            const issue_number = context.payload.issue.number;
+
+            // Read live labels (not the event payload) so labels added at creation
+            // time via the API or by another automation are seen — consistent with
+            // the live-state reads in the PR jobs above.
+            const current = (await github.paginate(github.rest.issues.listLabelsOnIssue, {
+              owner, repo, issue_number, per_page: 100,
+            })).map(l => l.name);
+            if (current.includes('needs-triage')) {
+              core.info('Issue already has needs-triage; nothing to do.');
+              return;
+            }
+            // Self-heal: create the label if it does not exist yet.
+            try {
+              await github.rest.issues.createLabel({
+                owner, repo, name: 'needs-triage', color: 'fef2c0',
+                description: 'Awaiting maintainer triage',
+              });
+            } catch (e) {
+              if (e.status !== 422) throw e; // 422 = already exists
+            }
+            await github.rest.issues.addLabels({
+              owner, repo, issue_number, labels: ['needs-triage'],
+            });
+            core.info(`Added needs-triage to #${issue_number}.`);

README.md

@@ -247,6 +247,9 @@ Access: http://localhost:2026
 
 The unified nginx endpoint is same-origin by default and does not emit browser CORS headers. If you run a split-origin or port-forwarded browser client, set `GATEWAY_CORS_ORIGINS` to comma-separated exact origins such as `http://localhost:3000`; the Gateway then applies the CORS allowlist and matching CSRF origin checks.
 
+> [!IMPORTANT]
+> The Gateway holds run state (RunManager and the stream bridge) in process, so production defaults to a single Gateway worker (`GATEWAY_WORKERS=1`). Raising the worker count without a shared cross-worker stream bridge — which is not yet available — breaks run cancellation, SSE reconnects, request de-duplication, and IM channels, because nginx uses no sticky sessions and each worker keeps its own run state. Scale a single worker up with more CPU/RAM (or move the database and sandbox onto dedicated tiers) instead of raising `GATEWAY_WORKERS`.
+
 See [CONTRIBUTING.md](CONTRIBUTING.md) for detailed Docker development guide.
 
 #### Option 2: Local Development
@@ -340,6 +343,8 @@ See the [MCP Server Guide](backend/docs/MCP_SERVER.md) for detailed instructions
 
 DeerFlow supports receiving tasks from messaging apps. Channels auto-start when configured — no public IP required for any of them.
 
+DeerFlow can also expose user-owned IM channel connections in the workspace UI. When `channel_connections` is enabled, logged-in users can bind Telegram, Slack, or Discord from the sidebar / Settings > Channels. It reuses the existing outbound `channels.*` transports, so no public IP or provider callback URL is required. Incoming IM messages then run under the connected DeerFlow user account. See [IM Channel Connections](backend/docs/IM_CHANNEL_CONNECTIONS.md) for setup and security notes.
+
 | Channel | Transport | Difficulty |
 |---------|-----------|------------|
 | Telegram | Bot API (long-polling) | Easy |
@@ -585,6 +590,8 @@ A standard Agent Skill is a structured capability module — a Markdown file tha
 
 Skills are loaded progressively — only when the task needs them, not all at once. This keeps the context window lean and makes DeerFlow work well even with token-sensitive models.
 
+Users can explicitly activate an enabled skill for a single turn by starting the request with `/skill-name`, for example `/data-analysis analyze uploads/foo.csv`. DeerFlow loads that skill's `SKILL.md` as hidden current-turn context while leaving the base prompt limited to skill metadata. Slash activation respects disabled skills, custom-agent skill whitelists, and existing channel commands such as `/new` and `/help`.
+
 When you install `.skill` archives through the Gateway, DeerFlow accepts standard optional frontmatter metadata such as `version`, `author`, and `compatibility` instead of rejecting otherwise valid external skills.
 
 Tools follow the same philosophy. DeerFlow comes with a core toolset — web search, web fetch, file operations, bash execution — and supports custom tools via MCP servers and Python functions. Swap anything. Add anything.

backend/.gitignore

@@ -24,5 +24,10 @@ config.yaml
 # Langgraph
 .langgraph_api
 
+# Sandbox runtime working dir — pre-created and excluded from uvicorn reload
+# (scripts/serve.sh, docker/dev-entrypoint.sh). Anchored so it does not match
+# the source package backend/packages/harness/deerflow/sandbox/.
+/sandbox/
+
 # Claude Code settings
 .claude/settings.local.json

backend/CLAUDE.md

@@ -192,7 +192,7 @@ from deerflow.config import get_app_config
 
 ### Middleware Chain
 
-Lead-agent middlewares are assembled in strict append order across `packages/harness/deerflow/agents/middlewares/tool_error_handling_middleware.py` (`build_lead_runtime_middlewares`) and `packages/harness/deerflow/agents/lead_agent/agent.py` (`_build_middlewares`):
+Lead-agent middlewares are assembled in strict append order across `packages/harness/deerflow/agents/middlewares/tool_error_handling_middleware.py` (`build_lead_runtime_middlewares`) and `packages/harness/deerflow/agents/lead_agent/agent.py` (`build_middlewares`):
 
 1. **ThreadDataMiddleware** - Creates per-thread directories under the user's isolation scope (`backend/.deer-flow/users/{user_id}/threads/{thread_id}/user-data/{workspace,uploads,outputs}`); resolves `user_id` via `get_effective_user_id()` (falls back to `"default"` in no-auth mode); Web UI thread deletion now follows LangGraph thread removal with Gateway cleanup of the local thread directory
 2. **UploadsMiddleware** - Tracks and injects newly uploaded files into conversation
@@ -202,16 +202,17 @@ Lead-agent middlewares are assembled in strict append order across `packages/har
 6. **GuardrailMiddleware** - Pre-tool-call authorization via pluggable `GuardrailProvider` protocol (optional, if `guardrails.enabled` in config). Evaluates each tool call and returns error ToolMessage on deny. Three provider options: built-in `AllowlistProvider` (zero deps), OAP policy providers (e.g. `aport-agent-guardrails`), or custom providers. See [docs/GUARDRAILS.md](docs/GUARDRAILS.md) for setup, usage, and how to implement a provider.
 7. **SandboxAuditMiddleware** - Audits sandboxed shell/file operations for security logging before tool execution continues
 8. **ToolErrorHandlingMiddleware** - Converts tool exceptions into error `ToolMessage`s so the run can continue instead of aborting
-9. **SummarizationMiddleware** - Context reduction when approaching token limits (optional, if enabled)
-10. **TodoListMiddleware** - Task tracking with `write_todos` tool (optional, if plan_mode)
-11. **TokenUsageMiddleware** - Records token usage metrics when token tracking is enabled (optional); subagent usage is cached by `tool_call_id` only while token usage is enabled and merged back into the dispatching AIMessage by message position rather than message id
-12. **TitleMiddleware** - Auto-generates thread title after first complete exchange and normalizes structured message content before prompting the title model
-13. **MemoryMiddleware** - Queues conversations for async memory update (filters to user + final AI responses)
-14. **ViewImageMiddleware** - Injects base64 image data before LLM call (conditional on vision support)
-15. **DeferredToolFilterMiddleware** - Hides deferred (MCP) tool schemas from the bound model using a build-time deferred-name set + catalog hash, reading per-thread promotions from `ThreadState.promoted` (hash-scoped, no ContextVar); a tool becomes bound on subsequent turns after `tool_search` returns its schema (optional, if `tool_search.enabled`)
-16. **SubagentLimitMiddleware** - Truncates excess `task` tool calls from model response to enforce `MAX_CONCURRENT_SUBAGENTS` limit (optional, if `subagent_enabled`)
-17. **LoopDetectionMiddleware** - Detects repeated tool-call loops; hard-stop responses clear both structured `tool_calls` and raw provider tool-call metadata before forcing a final text answer
-18. **ClarificationMiddleware** - Intercepts `ask_clarification` tool calls, interrupts via `Command(goto=END)` (must be last)
+9. **SkillActivationMiddleware** - Detects strict `/skill-name task` syntax on the latest real user message, resolves only enabled and runtime-allowed skills, reads `SKILL.md` from trusted skill storage, injects the skill body as hidden current-turn model context, and records a `middleware:skill_activation` audit event with skill name, category, path, and content hash
+10. **SummarizationMiddleware** - Context reduction when approaching token limits (optional, if enabled)
+11. **TodoListMiddleware** - Task tracking with `write_todos` tool (optional, if plan_mode)
+12. **TokenUsageMiddleware** - Records token usage metrics when token tracking is enabled (optional); subagent usage is cached by `tool_call_id` only while token usage is enabled and merged back into the dispatching AIMessage by message position rather than message id
+13. **TitleMiddleware** - Auto-generates thread title after first complete exchange and normalizes structured message content before prompting the title model
+14. **MemoryMiddleware** - Queues conversations for async memory update (filters to user + final AI responses)
+15. **ViewImageMiddleware** - Injects base64 image data before LLM call (conditional on vision support)
+16. **DeferredToolFilterMiddleware** - Hides deferred (MCP) tool schemas from the bound model using a build-time deferred-name set + catalog hash, reading per-thread promotions from `ThreadState.promoted` (hash-scoped, no ContextVar); a tool becomes bound on subsequent turns after `tool_search` returns its schema (optional, if `tool_search.enabled`)
+17. **SubagentLimitMiddleware** - Truncates excess `task` tool calls from model response to enforce `MAX_CONCURRENT_SUBAGENTS` limit (optional, if `subagent_enabled`)
+18. **LoopDetectionMiddleware** - Detects repeated tool-call loops; hard-stop responses clear both structured `tool_calls` and raw provider tool-call metadata before forcing a final text answer
+19. **ClarificationMiddleware** - Intercepts `ask_clarification` tool calls, interrupts via `Command(goto=END)` (must be last)
 
 ### Configuration System
 
@@ -263,7 +264,7 @@ CORS is same-origin by default when requests enter through nginx on port 2026. S
 | **Uploads** (`/api/threads/{id}/uploads`) | `POST /` - upload files (auto-converts PDF/PPT/Excel/Word); `GET /list` - list; `DELETE /{filename}` - delete |
 | **Threads** (`/api/threads/{id}`) | `DELETE /` - remove DeerFlow-managed local thread data after LangGraph thread deletion; unexpected failures are logged server-side and return a generic 500 detail |
 | **Artifacts** (`/api/threads/{id}/artifacts`) | `GET /{path}` - serve artifacts; active content types (`text/html`, `application/xhtml+xml`, `image/svg+xml`) are always forced as download attachments to reduce XSS risk; `?download=true` still forces download for other file types |
-| **Suggestions** (`/api/threads/{id}/suggestions`) | `POST /` - generate follow-up questions; rich list/block model content is normalized before JSON parsing |
+| **Suggestions** (`/api/threads/{id}/suggestions`) | `POST /` - generate follow-up questions; rich list/block model content is normalized and inline reasoning (`<think>...</think>`, including unclosed/truncated blocks from reasoning models like MiniMax-M3) is stripped before JSON parsing |
 | **Thread Runs** (`/api/threads/{id}/runs`) | `POST /` - create background run; `POST /stream` - create + SSE stream; `POST /wait` - create + block; `GET /` - list runs; `GET /{rid}` - run details; `POST /{rid}/cancel` - cancel; `GET /{rid}/join` - join SSE; `GET /{rid}/messages` - paginated messages `{data, has_more}`; `GET /{rid}/events` - full event stream; `GET /../messages` - thread messages with feedback; `GET /../token-usage` - aggregate tokens |
 | **Feedback** (`/api/threads/{id}/runs/{rid}/feedback`) | `PUT /` - upsert feedback; `DELETE /` - delete user feedback; `POST /` - create feedback; `GET /` - list feedback; `GET /stats` - aggregate stats; `DELETE /{fid}` - delete specific |
 | **Runs** (`/api/runs`) | `POST /stream` - stateless run + SSE; `POST /wait` - stateless run + block; `GET /{rid}/messages` - paginated messages by run_id `{data, has_more}` (cursor: `after_seq`/`before_seq`); `GET /{rid}/feedback` - list feedback by run_id |
@@ -305,6 +306,7 @@ Proxied through nginx: `/api/langgraph/*` → Gateway LangGraph-compatible runti
 **Concurrency**: `MAX_CONCURRENT_SUBAGENTS = 3` enforced by `SubagentLimitMiddleware` (truncates excess tool calls in `after_model`), 15-minute timeout
 **Flow**: `task()` tool → `SubagentExecutor` → background thread → poll 5s → SSE events → result
 **Events**: `task_started`, `task_running`, `task_completed`/`task_failed`/`task_timed_out`
+**Deferred MCP tools** (if `tool_search.enabled`): `SubagentExecutor._build_initial_state` assembles deferral after policy filtering via the shared `assemble_deferred_tools` (fail-closed), appends the `tool_search` tool, injects the `<available-deferred-tools>` section into the subagent's `SystemMessage`, and threads the setup to `_create_agent`, which attaches `DeferredToolFilterMiddleware` through `build_subagent_runtime_middlewares(deferred_setup=...)`. Subagents thus withhold full MCP schemas until promotion, same as the lead agent; each task run gets a fresh `ThreadState` so promotion is isolated per run
 
 ### Tool System (`packages/harness/deerflow/tools/`)
 
@@ -347,6 +349,7 @@ Proxied through nginx: `/api/langgraph/*` → Gateway LangGraph-compatible runti
 - **Format**: Directory with `SKILL.md` (YAML frontmatter: name, description, license, allowed-tools)
 - **Loading**: `load_skills()` recursively scans `skills/{public,custom}` for `SKILL.md`, parses metadata, and reads enabled state from extensions_config.json
 - **Injection**: Enabled skills listed in agent system prompt with container paths
+- **Slash activation**: `/skill-name task` loads that enabled skill's `SKILL.md` for the current model call only. The resolver rejects leading whitespace, missing separators, reserved channel commands (`/new`, `/help`, `/bootstrap`, `/status`, `/models`, `/memory`), disabled skills, and skills outside a custom agent's whitelist.
 - **Installation**: `POST /api/skills/install` extracts .skill ZIP archive to custom/ directory
 
 ### Model Factory (`packages/harness/deerflow/models/factory.py`)
@@ -366,8 +369,7 @@ Proxied through nginx: `/api/langgraph/*` → Gateway LangGraph-compatible runti
 
 ### IM Channels System (`app/channels/`)
 
-Bridges external messaging platforms (Feishu, Slack, Telegram, DingTalk) to the DeerFlow agent via Gateway's LangGraph-compatible API.
-
+Bridges external messaging platforms (Feishu, Slack, Telegram, Discord, DingTalk) to the DeerFlow agent via Gateway's LangGraph-compatible API.
 
 **Architecture**: Channels communicate with Gateway through the `langgraph-sdk` HTTP client (same as the frontend), ensuring threads are created and managed server-side. The internal SDK client injects process-local internal auth plus a matching CSRF cookie/header pair so Gateway accepts state-changing thread/run requests from channel workers without relying on browser session cookies.
 
@@ -377,18 +379,21 @@ Bridges external messaging platforms (Feishu, Slack, Telegram, DingTalk) to the
 - `manager.py` - Core dispatcher: creates threads via `client.threads.create()`, routes commands, keeps Slack/Telegram on `client.runs.wait()`, and uses `client.runs.stream(["messages-tuple", "values"])` for Feishu incremental outbound updates
 - `base.py` - Abstract `Channel` base class (start/stop/send lifecycle)
 - `service.py` - Manages lifecycle of all configured channels from `config.yaml`
-- `slack.py` / `feishu.py` / `telegram.py` / `dingtalk.py` - Platform-specific implementations (`feishu.py` tracks the running card `message_id` in memory and patches the same card in place; `dingtalk.py` optionally uses AI Card streaming for in-place updates when `card_template_id` is configured)
+- `slack.py` / `feishu.py` / `telegram.py` / `discord.py` / `dingtalk.py` - Platform-specific implementations (`feishu.py` tracks the running card `message_id` in memory and patches the same card in place; `dingtalk.py` optionally uses AI Card streaming for in-place updates when `card_template_id` is configured)
+- `app/gateway/routers/channel_connections.py` - Browser-facing user connection and disconnect APIs
+- `deerflow.persistence.channel_connections` - SQL-backed user-owned connection, optional credential, connect state, and conversation store
 
 **Message Flow**:
 1. External platform -> Channel impl -> `MessageBus.publish_inbound()`
 2. `ChannelManager._dispatch_loop()` consumes from queue
-3. For chat: look up/create thread through Gateway's LangGraph-compatible API
-4. Feishu chat: `runs.stream()` → accumulate AI text → publish multiple outbound updates (`is_final=False`) → publish final outbound (`is_final=True`)
-5. Slack/Telegram chat: `runs.wait()` → extract final response → publish outbound
-6. Feishu channel sends one running reply card up front, then patches the same card for each outbound update (card JSON sets `config.update_multi=true` for Feishu's patch API requirement)
-7. DingTalk AI Card mode (when `card_template_id` configured): `runs.stream()` → create card with initial text → stream updates via `PUT /v1.0/card/streaming` → finalize on `is_final=True`. Falls back to `sampleMarkdown` if card creation or streaming fails
-8. For commands (`/new`, `/status`, `/models`, `/memory`, `/help`): handle locally or query Gateway API
-9. Outbound → channel callbacks → platform reply
+3. For user-owned channel connections, incoming messages carry `connection_id`, `owner_user_id`, and `workspace_id`; `owner_user_id` becomes the DeerFlow run `user_id`, while the raw platform user id remains `channel_user_id`
+4. For chat: look up/create thread through Gateway's LangGraph-compatible API
+5. Feishu chat: `runs.stream()` → accumulate AI text → publish multiple outbound updates (`is_final=False`) → publish final outbound (`is_final=True`)
+6. Slack/Telegram chat: `runs.wait()` → extract final response → publish outbound
+7. Feishu channel sends one running reply card up front, then patches the same card for each outbound update (card JSON sets `config.update_multi=true` for Feishu's patch API requirement)
+8. DingTalk AI Card mode (when `card_template_id` configured): `runs.stream()` → create card with initial text → stream updates via `PUT /v1.0/card/streaming` → finalize on `is_final=True`. Falls back to `sampleMarkdown` if card creation or streaming fails
+9. For commands (`/new`, `/status`, `/models`, `/memory`, `/help`): handle locally or query Gateway API
+10. Outbound → channel callbacks → platform reply
 
 **Configuration** (`config.yaml` -> `channels`):
 - `langgraph_url` - LangGraph-compatible Gateway API base URL (default: `http://localhost:8001/api`)
@@ -396,6 +401,16 @@ Bridges external messaging platforms (Feishu, Slack, Telegram, DingTalk) to the
 - In Docker Compose, IM channels run inside the `gateway` container, so `localhost` points back to that container. Use `http://gateway:8001/api` for `langgraph_url` and `http://gateway:8001` for `gateway_url`, or set `DEER_FLOW_CHANNELS_LANGGRAPH_URL` / `DEER_FLOW_CHANNELS_GATEWAY_URL`.
 - Per-channel configs: `feishu` (app_id, app_secret), `slack` (bot_token, app_token), `telegram` (bot_token), `dingtalk` (client_id, client_secret, optional `card_template_id` for AI Card streaming)
 
+**User-owned channel connections** (`config.yaml` -> `channel_connections`):
+- Disabled by default. It is a user-binding layer on top of the existing `channels.*` runtime config, not a replacement for provider bot credentials.
+- No public IP, OAuth callback URL, or provider webhook route is required by the current implementation.
+- Telegram uses a deep-link `/start <code>` flow over the existing long-polling worker. Slack uses `/connect <code>` over the existing Socket Mode worker. Discord uses `/connect <code>` over the existing Gateway worker.
+- Frontend APIs: `GET /api/channels/providers`, `GET /api/channels/connections`, `POST /api/channels/{provider}/connect`, and `DELETE /api/channels/connections/{connection_id}`.
+- Browser APIs remain protected by normal Gateway auth/CSRF. Provider messages arrive through the already-configured channel workers.
+- Slack replies use the configured operator bot token from `channels.slack` unless a future provider-token flow stores per-connection credentials.
+- Telegram, Slack, and Discord workers resolve incoming platform identities to connection records before reaching `ChannelManager`.
+- See `backend/docs/IM_CHANNEL_CONNECTIONS.md` for provider setup and operational notes.
+
 
 ### Memory System (`packages/harness/deerflow/agents/memory/`)
 
@@ -492,7 +507,7 @@ Both can be modified at runtime via Gateway API endpoints or `DeerFlowClient` me
   - `"messages-tuple"` — per-chunk update: for AI text this is a **delta** (concat per `id` to rebuild the full message); tool calls and tool results are emitted once each
   - `"custom"` — forwarded from `StreamWriter`
   - `"end"` — stream finished (carries cumulative `usage` counted once per message id)
-- Agent created lazily via `create_agent()` + `_build_middlewares()`, same as `make_lead_agent`
+- Agent created lazily via `create_agent()` + `build_middlewares()`, same as `make_lead_agent`
 - Supports `checkpointer` parameter for state persistence across turns
 - `reset_agent()` forces agent recreation (e.g. after memory or skill changes)
 - See [docs/STREAMING.md](docs/STREAMING.md) for the full design: why Gateway and DeerFlowClient are parallel paths, LangGraph's `stream_mode` semantics, the per-id dedup invariants, and regression testing strategy

backend/app/channels/commands.py

@@ -18,3 +18,10 @@ KNOWN_CHANNEL_COMMANDS: frozenset[str] = frozenset(
         "/help",
     }
 )
+
+
+def is_known_channel_command(text: str) -> bool:
+    """Return whether text starts with a registered channel control command."""
+    if not text.startswith("/"):
+        return False
+    return text.split(maxsplit=1)[0].lower() in KNOWN_CHANNEL_COMMANDS

backend/app/channels/dingtalk.py

@@ -14,7 +14,7 @@ from typing import Any
 import httpx
 
 from app.channels.base import Channel
-from app.channels.commands import KNOWN_CHANNEL_COMMANDS
+from app.channels.commands import is_known_channel_command
 from app.channels.message_bus import InboundMessage, InboundMessageType, MessageBus, OutboundMessage, ResolvedAttachment
 
 logger = logging.getLogger(__name__)
@@ -59,9 +59,7 @@ def _normalize_allowed_users(allowed_users: Any) -> set[str]:
 
 
 def _is_dingtalk_command(text: str) -> bool:
-    if not text.startswith("/"):
-        return False
-    return text.split(maxsplit=1)[0].lower() in KNOWN_CHANNEL_COMMANDS
+    return is_known_channel_command(text)
 
 
 def _extract_text_from_rich_text(rich_text_list: list) -> str:

backend/app/channels/discord.py

@@ -10,13 +10,24 @@ from pathlib import Path
 from typing import Any
 
 from app.channels.base import Channel
-from app.channels.message_bus import InboundMessageType, MessageBus, OutboundMessage, ResolvedAttachment
+from app.channels.commands import is_known_channel_command
+from app.channels.message_bus import InboundMessage, InboundMessageType, MessageBus, OutboundMessage, ResolvedAttachment
 
 logger = logging.getLogger(__name__)
 
 _DISCORD_MAX_MESSAGE_LEN = 2000
 
 
+def _extract_connect_code(text: str) -> str | None:
+    parts = text.strip().split()
+    if len(parts) < 2:
+        return None
+    command = parts[0].lower()
+    if command in {"/connect", "connect"}:
+        return parts[1]
+    return None
+
+
 class DiscordChannel(Channel):
     """Discord bot channel.
 
@@ -69,6 +80,7 @@ class DiscordChannel(Channel):
         self._discord_loop: asyncio.AbstractEventLoop | None = None
         self._main_loop: asyncio.AbstractEventLoop | None = None
         self._discord_module = None
+        self._connection_repo = config.get("connection_repo")
 
     async def start(self) -> None:
         if self._running:
@@ -286,6 +298,10 @@ class DiscordChannel(Channel):
             text = text.replace(bot_mention or "", "").replace(alt_mention or "", "").replace(standard_mention or "", "").strip()
             # Don't return early if text is empty — still process the mention (e.g., create thread)
 
+        connect_code = _extract_connect_code(text)
+        if connect_code and await self._bind_connection_from_connect_code(message, connect_code):
+            return
+
         # --- Determine thread/channel routing and typing target ---
         thread_id = None
         chat_id = None
@@ -300,7 +316,7 @@ class DiscordChannel(Channel):
 
             # If this is a known active thread, process normally
             if thread_id in self._active_thread_ids:
-                msg_type = InboundMessageType.COMMAND if text.startswith("/") else InboundMessageType.CHAT
+                msg_type = InboundMessageType.COMMAND if is_known_channel_command(text) else InboundMessageType.CHAT
                 inbound = self._make_inbound(
                     chat_id=chat_id,
                     user_id=str(message.author.id),
@@ -314,6 +330,7 @@ class DiscordChannel(Channel):
                     },
                 )
                 inbound.topic_id = thread_id
+                inbound = await self._attach_connection_identity(inbound, guild_id=str(guild.id) if guild else None)
                 self._publish(inbound)
                 # Start typing indicator in the thread
                 if typing_target:
@@ -407,7 +424,7 @@ class DiscordChannel(Channel):
             chat_id = channel_id
             typing_target = message.channel  # Type into the channel
 
-        msg_type = InboundMessageType.COMMAND if text.startswith("/") else InboundMessageType.CHAT
+        msg_type = InboundMessageType.COMMAND if is_known_channel_command(text) else InboundMessageType.CHAT
         inbound = self._make_inbound(
             chat_id=chat_id,
             user_id=str(message.author.id),
@@ -421,6 +438,7 @@ class DiscordChannel(Channel):
             },
         )
         inbound.topic_id = thread_id
+        inbound = await self._attach_connection_identity(inbound, guild_id=str(guild.id) if guild else None)
 
         # Start typing indicator in the correct target (thread or channel)
         if typing_target:
@@ -435,6 +453,76 @@ class DiscordChannel(Channel):
             future = asyncio.run_coroutine_threadsafe(self.bus.publish_inbound(inbound), self._main_loop)
             future.add_done_callback(lambda f: logger.exception("[Discord] publish_inbound failed", exc_info=f.exception()) if f.exception() else None)
 
+    async def _attach_connection_identity(self, inbound: InboundMessage, guild_id: str | None = None) -> InboundMessage:
+        if self._connection_repo is None:
+            return inbound
+
+        connection = None
+        if guild_id:
+            connection = await self._connection_repo.find_connection_by_external_identity(
+                provider="discord",
+                external_account_id=inbound.user_id,
+                workspace_id=guild_id,
+            )
+        if connection is None:
+            connection = await self._connection_repo.find_connection_by_external_identity(
+                provider="discord",
+                external_account_id=inbound.user_id,
+                workspace_id=None,
+            )
+        if connection is None:
+            return inbound
+
+        inbound.connection_id = connection["id"]
+        inbound.owner_user_id = connection["owner_user_id"]
+        inbound.workspace_id = connection.get("workspace_id")
+        return inbound
+
+    async def _bind_connection_from_connect_code(self, message, code: str) -> bool:
+        if self._connection_repo is None or not code:
+            return False
+
+        state = await self._connection_repo.consume_oauth_state(provider="discord", state=code)
+        if state is None:
+            await self._send_connection_reply(message, "Discord connection code is invalid or expired.")
+            return True
+
+        guild = getattr(message, "guild", None)
+        channel = getattr(message, "channel", None)
+        author = getattr(message, "author", None)
+        user_id = str(getattr(author, "id", "") or "")
+        if not user_id:
+            await self._send_connection_reply(message, "Discord connection could not be completed from this message.")
+            return True
+
+        guild_id = str(getattr(guild, "id", "") or "") or None
+        await self._connection_repo.upsert_connection(
+            owner_user_id=state["owner_user_id"],
+            provider="discord",
+            external_account_id=user_id,
+            external_account_name=getattr(author, "display_name", None) or getattr(author, "name", None),
+            workspace_id=guild_id,
+            workspace_name=getattr(guild, "name", None) if guild is not None else None,
+            metadata={
+                "guild_id": guild_id,
+                "channel_id": str(getattr(channel, "id", "") or ""),
+            },
+            status="connected",
+        )
+        await self._send_connection_reply(message, "Discord connected to DeerFlow.")
+        return True
+
+    @staticmethod
+    async def _send_connection_reply(message, text: str) -> None:
+        channel = getattr(message, "channel", None)
+        send = getattr(channel, "send", None)
+        if send is None:
+            return
+        try:
+            await send(text)
+        except Exception:
+            logger.exception("[Discord] failed to send connection reply")
+
     def _run_client(self) -> None:
         self._discord_loop = asyncio.new_event_loop()
         asyncio.set_event_loop(self._discord_loop)

backend/app/channels/feishu.py

@@ -11,7 +11,7 @@ import time
 from typing import Any, Literal
 
 from app.channels.base import Channel
-from app.channels.commands import KNOWN_CHANNEL_COMMANDS
+from app.channels.commands import is_known_channel_command
 from app.channels.message_bus import (
     PENDING_CLARIFICATION_METADATA_KEY,
     RESOLVED_FROM_PENDING_CLARIFICATION_METADATA_KEY,
@@ -30,9 +30,7 @@ PENDING_CLARIFICATION_TTL_SECONDS = 30 * 60
 
 
 def _is_feishu_command(text: str) -> bool:
-    if not text.startswith("/"):
-        return False
-    return text.split(maxsplit=1)[0].lower() in KNOWN_CHANNEL_COMMANDS
+    return is_known_channel_command(text)
 
 
 class FeishuChannel(Channel):

backend/app/channels/manager.py

@@ -8,6 +8,7 @@ import mimetypes
 import re
 import time
 from collections.abc import Awaitable, Callable, Mapping
+from dataclasses import dataclass
 from pathlib import Path
 from typing import Any
 
@@ -26,8 +27,13 @@ from app.channels.message_bus import (
 from app.channels.store import ChannelStore
 from app.gateway.csrf_middleware import CSRF_COOKIE_NAME, CSRF_HEADER_NAME, generate_csrf_token
 from app.gateway.internal_auth import create_internal_auth_headers
+from deerflow.config.agents_config import load_agent_config
 from deerflow.config.paths import make_safe_user_id
 from deerflow.runtime.user_context import get_effective_user_id
+from deerflow.skills.slash import parse_slash_skill_reference
+from deerflow.skills.storage import get_or_new_skill_storage
+from deerflow.skills.storage.skill_storage import SkillStorage
+from deerflow.utils.messages import ORIGINAL_USER_CONTENT_KEY
 
 logger = logging.getLogger(__name__)
 
@@ -124,6 +130,16 @@ class InvalidChannelSessionConfigError(ValueError):
     """Raised when IM channel session overrides contain invalid agent config."""
 
 
+class SlashSkillCommandResolutionError(RuntimeError):
+    """Raised when IM slash-skill command resolution cannot complete safely."""
+
+
+@dataclass(frozen=True, slots=True)
+class _SlashSkillCommandResolution:
+    route_to_chat: bool = False
+    failure_message: str | None = None
+
+
 def _is_thread_busy_error(exc: BaseException | None) -> bool:
     if exc is None:
         return False
@@ -410,6 +426,46 @@ def _format_artifact_text(artifacts: list[str]) -> str:
 _OUTPUTS_VIRTUAL_PREFIX = "/mnt/user-data/outputs/"
 
 
+def _unknown_command_reply(command: str | None = None) -> str:
+    available = " | ".join(sorted(KNOWN_CHANNEL_COMMANDS))
+    if command:
+        return f"Unknown command: /{command}. Available commands: {available}"
+    return f"Unknown command. Available commands: {available}"
+
+
+def _human_input_message(content: str, *, original_content: str | None = None) -> dict[str, Any]:
+    message: dict[str, Any] = {"role": "human", "content": content}
+    if original_content is not None and original_content != content:
+        message["additional_kwargs"] = {ORIGINAL_USER_CONTENT_KEY: original_content}
+    return message
+
+
+def _resolve_slash_skill_command(
+    text: str,
+    available_skills: set[str] | None = None,
+    storage: SkillStorage | Callable[[], SkillStorage] | None = None,
+) -> _SlashSkillCommandResolution | None:
+    reference = parse_slash_skill_reference(text)
+    if reference is None:
+        return None
+    try:
+        resolved_storage = storage() if callable(storage) else storage or get_or_new_skill_storage()
+        skills = resolved_storage.load_skills(enabled_only=False)
+
+        skill = next((candidate for candidate in skills if candidate.name == reference.name), None)
+        if skill is None:
+            return None
+        if not skill.enabled:
+            return _SlashSkillCommandResolution(failure_message=f"Skill `/{reference.name}` is installed but disabled. Enable it before using slash activation.")
+        if available_skills is not None and reference.name not in available_skills:
+            return _SlashSkillCommandResolution(failure_message=f"Skill `/{reference.name}` is not available for this agent.")
+
+        return _SlashSkillCommandResolution(route_to_chat=True)
+    except Exception as exc:
+        logger.exception("[Manager] failed to resolve slash skill command")
+        raise SlashSkillCommandResolutionError("Failed to resolve slash skill command. Please check the skill configuration.") from exc
+
+
 def _resolve_attachments(thread_id: str, artifacts: list[str]) -> list[ResolvedAttachment]:
     """Resolve virtual artifact paths to host filesystem paths with metadata.
 
@@ -614,6 +670,7 @@ class ChannelManager:
         assistant_id: str = DEFAULT_ASSISTANT_ID,
         default_session: dict[str, Any] | None = None,
         channel_sessions: dict[str, Any] | None = None,
+        connection_repo: Any | None = None,
     ) -> None:
         self.bus = bus
         self.store = store
@@ -623,7 +680,9 @@ class ChannelManager:
         self._assistant_id = assistant_id
         self._default_session = _as_dict(default_session)
         self._channel_sessions = dict(channel_sessions or {})
+        self._connection_repo = connection_repo
         self._client = None  # lazy init — langgraph_sdk async client
+        self._skill_storage: SkillStorage | None = None
         self._csrf_token = generate_csrf_token()
         self._semaphore: asyncio.Semaphore | None = None
         self._running = False
@@ -671,12 +730,16 @@ class ChannelManager:
         configurable["checkpoint_ns"] = ""
         configurable["thread_id"] = thread_id
 
-        # ``user_id`` drives user-scoped filesystem buckets that only accept
-        # ``[A-Za-z0-9_-]``, so normalize the channel id and keep the raw value
-        # under ``channel_user_id`` for platform-facing lookups.
+        # ``user_id`` drives DeerFlow-owned memory, files, and thread buckets.
+        # For browser-connected IM channels, prefer the DeerFlow account that
+        # owns the connection. Preserve the raw platform user under
+        # ``channel_user_id`` for platform-facing lookups and audits.
         run_context_identity: dict[str, Any] = {"thread_id": thread_id}
-        if msg.user_id:
+        if msg.owner_user_id:
+            run_context_identity["user_id"] = make_safe_user_id(msg.owner_user_id)
+        elif msg.user_id:
             run_context_identity["user_id"] = make_safe_user_id(msg.user_id)
+        if msg.user_id:
             run_context_identity["channel_user_id"] = msg.user_id
 
         run_context = _merge_dicts(
@@ -696,6 +759,21 @@ class ChannelManager:
 
         return assistant_id, run_config, run_context
 
+    def _resolve_available_skill_names(self, msg: InboundMessage) -> set[str] | None:
+        thread_id = self.store.get_thread_id(msg.channel_name, msg.chat_id, topic_id=msg.topic_id) or ""
+        _, _, run_context = self._resolve_run_params(msg, thread_id)
+        if run_context.get("is_bootstrap"):
+            return {"bootstrap"}
+
+        agent_name = run_context.get("agent_name")
+        if not isinstance(agent_name, str) or not agent_name.strip():
+            return None
+
+        agent_config = load_agent_config(_normalize_custom_agent_name(agent_name))
+        if agent_config and agent_config.skills is not None:
+            return set(agent_config.skills)
+        return None
+
     # -- LangGraph SDK client (lazy) ----------------------------------------
 
     def _get_client(self):
@@ -713,6 +791,11 @@ class ChannelManager:
             )
         return self._client
 
+    def _get_skill_storage(self) -> SkillStorage:
+        if self._skill_storage is None:
+            self._skill_storage = get_or_new_skill_storage()
+        return self._skill_storage
+
     # -- lifecycle ---------------------------------------------------------
 
     async def start(self) -> None:
@@ -782,6 +865,14 @@ class ChannelManager:
                     exc,
                 )
                 await self._send_error(msg, str(exc))
+            except SlashSkillCommandResolutionError as exc:
+                logger.warning(
+                    "Slash skill command resolution failed for %s (chat=%s): %s",
+                    msg.channel_name,
+                    msg.chat_id,
+                    exc,
+                )
+                await self._send_error(msg, str(exc))
             except Exception:
                 logger.exception(
                     "Error handling message from %s (chat=%s)",
@@ -792,10 +883,27 @@ class ChannelManager:
 
     # -- chat handling -----------------------------------------------------
 
-    async def _create_thread(self, client, msg: InboundMessage) -> str:
-        """Create a new thread through Gateway and store the mapping."""
-        thread = await client.threads.create()
-        thread_id = thread["thread_id"]
+    async def _lookup_thread_id(self, msg: InboundMessage) -> str | None:
+        if msg.connection_id and self._connection_repo is not None:
+            return await self._connection_repo.get_thread_id(
+                msg.connection_id,
+                msg.chat_id,
+                msg.topic_id,
+            )
+        return self.store.get_thread_id(msg.channel_name, msg.chat_id, topic_id=msg.topic_id)
+
+    async def _store_thread_id(self, msg: InboundMessage, thread_id: str) -> None:
+        if msg.connection_id and msg.owner_user_id and self._connection_repo is not None:
+            await self._connection_repo.set_thread_id(
+                connection_id=msg.connection_id,
+                owner_user_id=msg.owner_user_id,
+                provider=msg.channel_name,
+                external_conversation_id=msg.chat_id,
+                external_topic_id=msg.topic_id,
+                thread_id=thread_id,
+            )
+            return
+
         self.store.set_thread_id(
             msg.channel_name,
             msg.chat_id,
@@ -803,6 +911,12 @@ class ChannelManager:
             topic_id=msg.topic_id,
             user_id=msg.user_id,
         )
+
+    async def _create_thread(self, client, msg: InboundMessage) -> str:
+        """Create a new thread through Gateway and store the mapping."""
+        thread = await client.threads.create()
+        thread_id = thread["thread_id"]
+        await self._store_thread_id(msg, thread_id)
         logger.info("[Manager] new thread created through Gateway: thread_id=%s for chat_id=%s topic_id=%s", thread_id, msg.chat_id, msg.topic_id)
         return thread_id
 
@@ -812,7 +926,7 @@ class ChannelManager:
         # Look up existing DeerFlow thread.
         # topic_id may be None (e.g. Telegram private chats) — the store
         # handles this by using the "channel:chat_id" key without a topic suffix.
-        thread_id = self.store.get_thread_id(msg.channel_name, msg.chat_id, topic_id=msg.topic_id)
+        thread_id = await self._lookup_thread_id(msg)
         if thread_id:
             logger.info("[Manager] reusing thread: thread_id=%s for topic_id=%s", thread_id, msg.topic_id)
 
@@ -836,9 +950,11 @@ class ChannelManager:
         if extra_context:
             run_context.update(extra_context)
 
+        original_text = msg.text
         uploaded = await _ingest_inbound_files(thread_id, msg)
         if uploaded:
             msg.text = f"{_format_uploaded_files_block(uploaded)}\n\n{msg.text}".strip()
+        human_message = _human_input_message(msg.text, original_content=original_text)
 
         if self._channel_supports_streaming(msg.channel_name):
             await self._handle_streaming_chat(
@@ -848,6 +964,7 @@ class ChannelManager:
                 assistant_id,
                 run_config,
                 run_context,
+                human_message,
             )
             return
 
@@ -856,7 +973,7 @@ class ChannelManager:
             result = await client.runs.wait(
                 thread_id,
                 assistant_id,
-                input={"messages": [{"role": "human", "content": msg.text}]},
+                input={"messages": [human_message]},
                 config=run_config,
                 context=run_context,
                 multitask_strategy="reject",
@@ -896,6 +1013,8 @@ class ChannelManager:
             artifacts=artifacts,
             attachments=attachments,
             thread_ts=msg.thread_ts,
+            connection_id=msg.connection_id,
+            owner_user_id=msg.owner_user_id,
             metadata=_response_metadata(msg.metadata, pending_clarification=pending_clarification),
         )
         logger.info("[Manager] publishing outbound message to bus: channel=%s, chat_id=%s", msg.channel_name, msg.chat_id)
@@ -909,6 +1028,7 @@ class ChannelManager:
         assistant_id: str,
         run_config: dict[str, Any],
         run_context: dict[str, Any],
+        human_message: dict[str, Any],
     ) -> None:
         logger.info("[Manager] invoking runs.stream(thread_id=%s, text=%r)", thread_id, msg.text[:100])
 
@@ -924,7 +1044,7 @@ class ChannelManager:
             async for chunk in client.runs.stream(
                 thread_id,
                 assistant_id,
-                input={"messages": [{"role": "human", "content": msg.text}]},
+                input={"messages": [human_message]},
                 config=run_config,
                 context=run_context,
                 stream_mode=["messages-tuple", "values"],
@@ -958,6 +1078,8 @@ class ChannelManager:
                         text=latest_text,
                         is_final=False,
                         thread_ts=msg.thread_ts,
+                        connection_id=msg.connection_id,
+                        owner_user_id=msg.owner_user_id,
                         metadata=_response_metadata(msg.metadata),
                     )
                 )
@@ -1004,6 +1126,8 @@ class ChannelManager:
                     attachments=attachments,
                     is_final=True,
                     thread_ts=msg.thread_ts,
+                    connection_id=msg.connection_id,
+                    owner_user_id=msg.owner_user_id,
                     metadata=_response_metadata(msg.metadata, pending_clarification=pending_clarification),
                 )
             )
@@ -1011,11 +1135,20 @@ class ChannelManager:
     # -- command handling --------------------------------------------------
 
     async def _handle_command(self, msg: InboundMessage) -> None:
-        text = msg.text.strip()
+        raw_text = msg.text
+        text = raw_text.strip()
         parts = text.split(maxsplit=1)
-        command = parts[0].lower().lstrip("/")
+        reply: str | None = None
+        if not parts:
+            command = None
+            reply = _unknown_command_reply()
+        else:
+            command = parts[0].lower().removeprefix("/")
 
-        if command == "bootstrap":
+        if reply is None and not raw_text.startswith("/"):
+            reply = _unknown_command_reply(command)
+
+        if reply is None and command == "bootstrap":
             from dataclasses import replace as _dc_replace
 
             chat_text = parts[1] if len(parts) > 1 else "Initialize workspace"
@@ -1023,27 +1156,21 @@ class ChannelManager:
             await self._handle_chat(chat_msg, extra_context={"is_bootstrap": True})
             return
 
-        if command == "new":
+        if reply is None and command == "new":
             # Create a new thread through Gateway
             client = self._get_client()
             thread = await client.threads.create()
             new_thread_id = thread["thread_id"]
-            self.store.set_thread_id(
-                msg.channel_name,
-                msg.chat_id,
-                new_thread_id,
-                topic_id=msg.topic_id,
-                user_id=msg.user_id,
-            )
+            await self._store_thread_id(msg, new_thread_id)
             reply = "New conversation started."
-        elif command == "status":
-            thread_id = self.store.get_thread_id(msg.channel_name, msg.chat_id, topic_id=msg.topic_id)
+        elif reply is None and command == "status":
+            thread_id = await self._lookup_thread_id(msg)
             reply = f"Active thread: {thread_id}" if thread_id else "No active conversation."
-        elif command == "models":
+        elif reply is None and command == "models":
             reply = await self._fetch_gateway("/api/models", "models")
-        elif command == "memory":
+        elif reply is None and command == "memory":
             reply = await self._fetch_gateway("/api/memory", "memory")
-        elif command == "help":
+        elif reply is None and command == "help":
             reply = (
                 "Available commands:\n"
                 "/bootstrap — Start a bootstrap session (enables agent setup)\n"
@@ -1051,18 +1178,36 @@ class ChannelManager:
                 "/status — Show current thread info\n"
                 "/models — List available models\n"
                 "/memory — Show memory status\n"
+                "/<skill-name> <task> — Activate an enabled skill for one turn\n"
                 "/help — Show this help"
             )
-        else:
-            available = " | ".join(sorted(KNOWN_CHANNEL_COMMANDS))
-            reply = f"Unknown command: /{command}. Available commands: {available}"
+        elif reply is None:
+            slash_resolution = await asyncio.to_thread(
+                lambda: _resolve_slash_skill_command(
+                    raw_text,
+                    self._resolve_available_skill_names(msg),
+                    self._get_skill_storage,
+                )
+            )
+            if slash_resolution and slash_resolution.failure_message:
+                reply = slash_resolution.failure_message
+            elif slash_resolution and slash_resolution.route_to_chat:
+                from dataclasses import replace as _dc_replace
+
+                chat_msg = _dc_replace(msg, msg_type=InboundMessageType.CHAT)
+                await self._handle_chat(chat_msg)
+                return
+            else:
+                reply = _unknown_command_reply(command)
 
         outbound = OutboundMessage(
             channel_name=msg.channel_name,
             chat_id=msg.chat_id,
-            thread_id=self.store.get_thread_id(msg.channel_name, msg.chat_id) or "",
+            thread_id=await self._lookup_thread_id(msg) or "",
             text=reply,
             thread_ts=msg.thread_ts,
+            connection_id=msg.connection_id,
+            owner_user_id=msg.owner_user_id,
             metadata=_slim_metadata(msg.metadata),
         )
         await self.bus.publish_outbound(outbound)
@@ -1098,9 +1243,11 @@ class ChannelManager:
         outbound = OutboundMessage(
             channel_name=msg.channel_name,
             chat_id=msg.chat_id,
-            thread_id=self.store.get_thread_id(msg.channel_name, msg.chat_id) or "",
+            thread_id=await self._lookup_thread_id(msg) or "",
             text=error_text,
             thread_ts=msg.thread_ts,
+            connection_id=msg.connection_id,
+            owner_user_id=msg.owner_user_id,
             metadata=_slim_metadata(msg.metadata),
         )
         await self.bus.publish_outbound(outbound)

backend/app/channels/message_bus.py

@@ -44,6 +44,12 @@ class InboundMessage:
             Messages sharing the same ``topic_id`` within a ``chat_id`` will
             reuse the same DeerFlow thread.  When ``None``, each message
             creates a new thread (one-shot Q&A).
+        connection_id: Optional DeerFlow channel connection id. When present,
+            conversation mapping is scoped by the connection instead of the
+            legacy global ``channel_name:chat_id[:topic_id]`` key.
+        owner_user_id: DeerFlow user id that owns the channel connection.
+            Platform user ids stay in ``user_id``.
+        workspace_id: Optional external workspace/guild/team id.
         files: Optional list of file attachments (platform-specific dicts).
         metadata: Arbitrary extra data from the channel.
         created_at: Unix timestamp when the message was created.
@@ -56,6 +62,9 @@ class InboundMessage:
     msg_type: InboundMessageType = InboundMessageType.CHAT
     thread_ts: str | None = None
     topic_id: str | None = None
+    connection_id: str | None = None
+    owner_user_id: str | None = None
+    workspace_id: str | None = None
     files: list[dict[str, Any]] = field(default_factory=list)
     metadata: dict[str, Any] = field(default_factory=dict)
     created_at: float = field(default_factory=time.time)
@@ -95,6 +104,9 @@ class OutboundMessage:
         is_final: Whether this is the final message in the response stream.
         thread_ts: Optional platform thread identifier for threaded replies.
         metadata: Arbitrary extra data.
+        connection_id: Optional DeerFlow channel connection id used for
+            connection-specific outbound credentials.
+        owner_user_id: DeerFlow user id that owns the channel connection.
         created_at: Unix timestamp.
     """
 
@@ -106,6 +118,8 @@ class OutboundMessage:
     attachments: list[ResolvedAttachment] = field(default_factory=list)
     is_final: bool = True
     thread_ts: str | None = None
+    connection_id: str | None = None
+    owner_user_id: str | None = None
     metadata: dict[str, Any] = field(default_factory=dict)
     created_at: float = field(default_factory=time.time)
 

backend/app/channels/service.py

@@ -52,6 +52,31 @@ def _resolve_service_url(config: dict[str, Any], config_key: str, env_key: str,
     return default
 
 
+def _merge_channel_connection_runtime_config(channels_config: dict[str, Any], app_config: AppConfig) -> None:
+    connection_config = getattr(app_config, "channel_connections", None)
+    if connection_config is None or not getattr(connection_config, "enabled", False):
+        return
+
+
+def _make_connection_repo(app_config: AppConfig):
+    connection_config = getattr(app_config, "channel_connections", None)
+    if connection_config is None or not getattr(connection_config, "enabled", False):
+        return None
+
+    try:
+        from deerflow.persistence.channel_connections import ChannelConnectionRepository
+        from deerflow.persistence.engine import get_session_factory
+    except Exception:
+        logger.exception("Failed to import channel connection repository")
+        return None
+
+    session_factory = get_session_factory()
+    if session_factory is None:
+        logger.warning("Channel connections are enabled but database persistence is not available")
+        return None
+    return ChannelConnectionRepository(session_factory)
+
+
 class ChannelService:
     """Manages the lifecycle of all configured IM channels.
 
@@ -59,9 +84,10 @@ class ChannelService:
     instantiates enabled channels, and starts the ChannelManager dispatcher.
     """
 
-    def __init__(self, channels_config: dict[str, Any] | None = None) -> None:
+    def __init__(self, channels_config: dict[str, Any] | None = None, *, connection_repo: Any | None = None) -> None:
         self.bus = MessageBus()
         self.store = ChannelStore()
+        self._connection_repo = connection_repo
         config = dict(channels_config or {})
         langgraph_url = _resolve_service_url(config, "langgraph_url", _CHANNELS_LANGGRAPH_URL_ENV, DEFAULT_LANGGRAPH_URL)
         gateway_url = _resolve_service_url(config, "gateway_url", _CHANNELS_GATEWAY_URL_ENV, DEFAULT_GATEWAY_URL)
@@ -74,6 +100,7 @@ class ChannelService:
             gateway_url=gateway_url,
             default_session=default_session if isinstance(default_session, dict) else None,
             channel_sessions=channel_sessions,
+            connection_repo=connection_repo,
         )
         self._channels: dict[str, Any] = {}  # name -> Channel instance
         self._config = config
@@ -90,8 +117,9 @@ class ChannelService:
         # extra fields are allowed by AppConfig (extra="allow")
         extra = app_config.model_extra or {}
         if "channels" in extra:
-            channels_config = extra["channels"]
-        return cls(channels_config=channels_config)
+            channels_config = dict(extra["channels"] or {})
+        _merge_channel_connection_runtime_config(channels_config, app_config)
+        return cls(channels_config=channels_config, connection_repo=_make_connection_repo(app_config))
 
     async def start(self) -> None:
         """Start the manager and all enabled channels."""
@@ -169,6 +197,8 @@ class ChannelService:
         try:
             config = dict(config)
             config["channel_store"] = self.store
+            if self._connection_repo is not None:
+                config["connection_repo"] = self._connection_repo
             channel = channel_cls(bus=self.bus, config=config)
             self._channels[name] = channel
             await channel.start()

backend/app/channels/slack.py

@@ -9,6 +9,7 @@ from typing import Any
 from markdown_to_mrkdwn import SlackMarkdownConverter
 
 from app.channels.base import Channel
+from app.channels.commands import is_known_channel_command
 from app.channels.message_bus import InboundMessageType, MessageBus, OutboundMessage, ResolvedAttachment
 
 logger = logging.getLogger(__name__)
@@ -32,6 +33,30 @@ def _normalize_allowed_users(allowed_users: Any) -> set[str]:
     return {str(user_id) for user_id in values if str(user_id)}
 
 
+def _strip_leading_slack_bot_mention(text: str, bot_user_id: str | None) -> str:
+    if not bot_user_id:
+        return text
+    if not text.startswith("<@"):
+        return text
+    end = text.find(">")
+    if end <= 2:
+        return text
+    mentioned_user_id = text[2:end].split("|", 1)[0].lstrip("!")
+    if mentioned_user_id != bot_user_id:
+        return text
+    return text[end + 1 :].lstrip()
+
+
+def _extract_connect_code(text: str) -> str | None:
+    parts = text.strip().split()
+    if len(parts) < 2:
+        return None
+    command = parts[0].lower()
+    if command in {"/connect", "connect"}:
+        return parts[1]
+    return None
+
+
 class SlackChannel(Channel):
     """Slack IM channel using Socket Mode (WebSocket, no public IP).
 
@@ -49,6 +74,10 @@ class SlackChannel(Channel):
         self._web_client = None
         self._loop: asyncio.AbstractEventLoop | None = None
         self._allowed_users = _normalize_allowed_users(config.get("allowed_users", []))
+        self._connection_repo = config.get("connection_repo")
+        self._web_client_factory = config.get("web_client_factory")
+        configured_bot_user_id = config.get("bot_user_id")
+        self._bot_user_id = str(configured_bot_user_id).lstrip("@") if configured_bot_user_id else None
 
     async def start(self) -> None:
         if self._running:
@@ -63,15 +92,35 @@ class SlackChannel(Channel):
             return
 
         self._SocketModeResponse = SocketModeResponse
+        if self._web_client_factory is None:
+            self._web_client_factory = WebClient
 
         bot_token = self.config.get("bot_token", "")
         app_token = self.config.get("app_token", "")
 
+        if self._connection_repo is not None and self.config.get("event_delivery") == "http":
+            self._loop = asyncio.get_event_loop()
+            self._running = True
+            self.bus.subscribe_outbound(self._on_outbound)
+            logger.info("Slack channel started in HTTP Events mode")
+            return
+
         if not bot_token or not app_token:
             logger.error("Slack channel requires bot_token and app_token")
             return
 
-        self._web_client = WebClient(token=bot_token)
+        self._web_client = self._web_client_factory(token=bot_token)
+        if self._bot_user_id is None:
+            try:
+                auth_info = await asyncio.to_thread(self._web_client.auth_test)
+                user_id = auth_info.get("user_id") if isinstance(auth_info, dict) else None
+                if user_id is None:
+                    auth_get = getattr(auth_info, "get", None)
+                    user_id = auth_get("user_id") if callable(auth_get) else None
+                if isinstance(user_id, str) and user_id:
+                    self._bot_user_id = user_id
+            except Exception:
+                logger.warning("[Slack] failed to resolve bot user id; app mention text may include the bot mention", exc_info=True)
         self._socket_client = SocketModeClient(
             app_token=app_token,
             web_client=self._web_client,
@@ -96,7 +145,8 @@ class SlackChannel(Channel):
         logger.info("Slack channel stopped")
 
     async def send(self, msg: OutboundMessage, *, _max_retries: int = 3) -> None:
-        if not self._web_client:
+        web_client = await self._get_web_client_for_message(msg)
+        if not web_client:
             return
 
         kwargs: dict[str, Any] = {
@@ -109,11 +159,12 @@ class SlackChannel(Channel):
         last_exc: Exception | None = None
         for attempt in range(_max_retries):
             try:
-                await asyncio.to_thread(self._web_client.chat_postMessage, **kwargs)
+                await asyncio.to_thread(web_client.chat_postMessage, **kwargs)
                 # Add a completion reaction to the thread root
                 if msg.thread_ts:
                     await asyncio.to_thread(
-                        self._add_reaction,
+                        self._add_reaction_with_client,
+                        web_client,
                         msg.chat_id,
                         msg.thread_ts,
                         "white_check_mark",
@@ -137,7 +188,8 @@ class SlackChannel(Channel):
         if msg.thread_ts:
             try:
                 await asyncio.to_thread(
-                    self._add_reaction,
+                    self._add_reaction_with_client,
+                    web_client,
                     msg.chat_id,
                     msg.thread_ts,
                     "x",
@@ -149,7 +201,8 @@ class SlackChannel(Channel):
         raise last_exc
 
     async def send_file(self, msg: OutboundMessage, attachment: ResolvedAttachment) -> bool:
-        if not self._web_client:
+        web_client = await self._get_web_client_for_message(msg)
+        if not web_client:
             return False
 
         try:
@@ -162,7 +215,7 @@ class SlackChannel(Channel):
             if msg.thread_ts:
                 kwargs["thread_ts"] = msg.thread_ts
 
-            await asyncio.to_thread(self._web_client.files_upload_v2, **kwargs)
+            await asyncio.to_thread(web_client.files_upload_v2, **kwargs)
             logger.info("[Slack] file uploaded: %s to channel=%s", attachment.filename, msg.chat_id)
             return True
         except Exception:
@@ -171,12 +224,23 @@ class SlackChannel(Channel):
 
     # -- internal ----------------------------------------------------------
 
-    def _add_reaction(self, channel_id: str, timestamp: str, emoji: str) -> None:
-        """Add an emoji reaction to a message (best-effort, non-blocking)."""
-        if not self._web_client:
-            return
+    async def _get_web_client_for_message(self, msg: OutboundMessage):
+        if msg.connection_id and self._connection_repo is not None:
+            credentials = await self._connection_repo.get_credentials(msg.connection_id)
+            access_token = credentials.get("access_token") if credentials else None
+            if not access_token:
+                return self._web_client
+            if self._web_client_factory is None:
+                from slack_sdk import WebClient
+
+                self._web_client_factory = WebClient
+            return self._web_client_factory(token=access_token)
+        return self._web_client
+
+    @staticmethod
+    def _add_reaction_with_client(web_client, channel_id: str, timestamp: str, emoji: str) -> None:
         try:
-            self._web_client.reactions_add(
+            web_client.reactions_add(
                 channel=channel_id,
                 timestamp=timestamp,
                 name=emoji,
@@ -185,6 +249,12 @@ class SlackChannel(Channel):
             if "already_reacted" not in str(exc):
                 logger.warning("[Slack] failed to add reaction %s: %s", emoji, exc)
 
+    def _add_reaction(self, channel_id: str, timestamp: str, emoji: str) -> None:
+        """Add an emoji reaction to a message (best-effort, non-blocking)."""
+        if not self._web_client:
+            return
+        self._add_reaction_with_client(self._web_client, channel_id, timestamp, emoji)
+
     def _send_running_reply(self, channel_id: str, thread_ts: str) -> None:
         """Send a 'Working on it......' reply in the thread (called from SDK thread)."""
         if not self._web_client:
@@ -210,17 +280,26 @@ class SlackChannel(Channel):
             if event_type != "events_api":
                 return
 
+            if self._bot_user_id is None:
+                authorization = next((item for item in req.payload.get("authorizations", []) if isinstance(item, dict)), None)
+                user_id = authorization.get("user_id") if authorization else None
+                if isinstance(user_id, str) and user_id:
+                    self._bot_user_id = user_id
+
             event = req.payload.get("event", {})
             etype = event.get("type", "")
 
             # Handle message events (DM or @mention)
             if etype in ("message", "app_mention"):
-                self._handle_message_event(event)
+                self._handle_message_event(
+                    event,
+                    team_id=req.payload.get("team_id") or req.payload.get("team") or event.get("team"),
+                )
 
         except Exception:
             logger.exception("Error processing Slack event")
 
-    def _handle_message_event(self, event: dict) -> None:
+    def _handle_message_event(self, event: dict, *, team_id: str | None = None) -> None:
         # Ignore bot messages
         if event.get("bot_id") or event.get("subtype"):
             return
@@ -233,13 +312,28 @@ class SlackChannel(Channel):
             return
 
         text = event.get("text", "").strip()
+        if event.get("type") == "app_mention":
+            text = _strip_leading_slack_bot_mention(text, self._bot_user_id)
         if not text:
             return
 
+        connect_code = _extract_connect_code(text)
+        if connect_code:
+            if self._loop and self._loop.is_running():
+                asyncio.run_coroutine_threadsafe(
+                    self._bind_connection_from_connect_code(
+                        event=event,
+                        team_id=str(team_id or event.get("team") or ""),
+                        code=connect_code,
+                    ),
+                    self._loop,
+                )
+            return
+
         channel_id = event.get("channel", "")
         thread_ts = event.get("thread_ts") or event.get("ts", "")
 
-        if text.startswith("/"):
+        if is_known_channel_command(text):
             msg_type = InboundMessageType.COMMAND
         else:
             msg_type = InboundMessageType.CHAT
@@ -261,4 +355,73 @@ class SlackChannel(Channel):
             self._add_reaction(channel_id, event.get("ts", thread_ts), "eyes")
             # Send "running" reply first (fire-and-forget from SDK thread)
             self._send_running_reply(channel_id, thread_ts)
-            asyncio.run_coroutine_threadsafe(self.bus.publish_inbound(inbound), self._loop)
+            if self._connection_repo is None:
+                asyncio.run_coroutine_threadsafe(self.bus.publish_inbound(inbound), self._loop)
+            else:
+                asyncio.run_coroutine_threadsafe(self._publish_inbound_with_connection(inbound, team_id=team_id), self._loop)
+
+    async def _publish_inbound_with_connection(self, inbound, *, team_id: str | None = None) -> None:
+        inbound = await self._attach_connection_identity(inbound, team_id=team_id)
+        await self.bus.publish_inbound(inbound)
+
+    async def _attach_connection_identity(self, inbound, *, team_id: str | None = None):
+        if self._connection_repo is None:
+            return inbound
+
+        workspace_id = str(team_id or inbound.metadata.get("team_id") or "")
+        if not workspace_id:
+            return inbound
+
+        connection = await self._connection_repo.find_connection_by_external_identity(
+            provider="slack",
+            external_account_id=inbound.user_id,
+            workspace_id=workspace_id,
+        )
+        if connection is None:
+            return inbound
+
+        inbound.connection_id = connection["id"]
+        inbound.owner_user_id = connection["owner_user_id"]
+        inbound.workspace_id = connection.get("workspace_id")
+        return inbound
+
+    async def _bind_connection_from_connect_code(self, *, event: dict, team_id: str, code: str) -> bool:
+        if self._connection_repo is None or not code:
+            return False
+
+        channel_id = str(event.get("channel") or "")
+        thread_ts = str(event.get("thread_ts") or event.get("ts") or "")
+        state = await self._connection_repo.consume_oauth_state(provider="slack", state=code)
+        if state is None:
+            self._post_connection_reply(channel_id, "Slack connection code is invalid or expired.", thread_ts)
+            return True
+
+        user_id = str(event.get("user") or "")
+        if not user_id or not team_id:
+            self._post_connection_reply(channel_id, "Slack connection could not be completed from this message.", thread_ts)
+            return True
+
+        await self._connection_repo.upsert_connection(
+            owner_user_id=state["owner_user_id"],
+            provider="slack",
+            external_account_id=user_id,
+            workspace_id=team_id,
+            metadata={
+                "team_id": team_id,
+                "channel_id": channel_id,
+            },
+            status="connected",
+        )
+        self._post_connection_reply(channel_id, "Slack connected to DeerFlow.", thread_ts)
+        return True
+
+    def _post_connection_reply(self, channel_id: str, text: str, thread_ts: str | None = None) -> None:
+        if not self._web_client or not channel_id:
+            return
+        kwargs: dict[str, Any] = {"channel": channel_id, "text": text}
+        if thread_ts:
+            kwargs["thread_ts"] = thread_ts
+        try:
+            self._web_client.chat_postMessage(**kwargs)
+        except Exception:
+            logger.exception("[Slack] failed to send connection reply in channel=%s", channel_id)

backend/app/channels/telegram.py

@@ -35,6 +35,7 @@ class TelegramChannel(Channel):
                 pass
         # chat_id -> last sent message_id for threaded replies
         self._last_bot_message: dict[str, int] = {}
+        self._connection_repo = config.get("connection_repo")
 
     async def start(self) -> None:
         if self._running:
@@ -60,12 +61,17 @@ class TelegramChannel(Channel):
 
         # Command handlers
         app.add_handler(CommandHandler("start", self._cmd_start))
+        app.add_handler(CommandHandler("bootstrap", self._cmd_generic))
         app.add_handler(CommandHandler("new", self._cmd_generic))
         app.add_handler(CommandHandler("status", self._cmd_generic))
         app.add_handler(CommandHandler("models", self._cmd_generic))
         app.add_handler(CommandHandler("memory", self._cmd_generic))
         app.add_handler(CommandHandler("help", self._cmd_generic))
 
+        # Slash skill commands are dynamic and cannot all be pre-registered
+        # with Telegram, so route unknown slash commands through chat handling.
+        app.add_handler(MessageHandler(filters.TEXT & filters.COMMAND, self._on_text))
+
         # General message handler
         app.add_handler(MessageHandler(filters.TEXT & ~filters.COMMAND, self._on_text))
 
@@ -171,6 +177,26 @@ class TelegramChannel(Channel):
             logger.exception("[Telegram] failed to send file: %s", attachment.filename)
             return False
 
+    async def process_webhook_update(self, payload: dict[str, Any]) -> bool:
+        if not self._application:
+            return False
+        try:
+            from telegram import Update
+        except ImportError:
+            logger.error("python-telegram-bot is not installed. Install it with: uv add python-telegram-bot")
+            return False
+
+        update = Update.de_json(payload, self._application.bot)
+        if update is None:
+            return False
+
+        if self._tg_loop and self._tg_loop.is_running():
+            future = asyncio.run_coroutine_threadsafe(self._application.process_update(update), self._tg_loop)
+            await asyncio.wrap_future(future)
+        else:
+            await self._application.process_update(update)
+        return True
+
     # -- helpers -----------------------------------------------------------
 
     async def _send_running_reply(self, chat_id: str, reply_to_message_id: int) -> None:
@@ -228,10 +254,99 @@ class TelegramChannel(Channel):
             return True
         return user_id in self._allowed_users
 
+    @staticmethod
+    def _telegram_display_name(user) -> str:
+        full_name = getattr(user, "full_name", None)
+        if isinstance(full_name, str) and full_name:
+            return full_name
+        username = getattr(user, "username", None)
+        if isinstance(username, str) and username:
+            return username
+        return str(getattr(user, "id", ""))
+
+    async def _bind_connection_from_start_token(self, update, state_token: str) -> bool:
+        if self._connection_repo is None or not state_token:
+            return False
+
+        state = await self._connection_repo.consume_oauth_state(provider="telegram", state=state_token)
+        if state is None:
+            await update.message.reply_text("Telegram connection link is invalid or expired.")
+            return True
+
+        owner_user_id = state["owner_user_id"]
+        user_id = str(update.effective_user.id)
+        chat_id = str(update.effective_chat.id)
+        connection = await self._connection_repo.upsert_connection(
+            owner_user_id=owner_user_id,
+            provider="telegram",
+            external_account_id=user_id,
+            external_account_name=self._telegram_display_name(update.effective_user),
+            workspace_id=chat_id,
+            workspace_name=None,
+            metadata={
+                "chat_id": chat_id,
+                "chat_type": update.effective_chat.type,
+                "telegram_username": getattr(update.effective_user, "username", None),
+            },
+            status="connected",
+        )
+        logger.info("[Telegram] bound chat=%s user=%s to DeerFlow user=%s connection=%s", chat_id, user_id, owner_user_id, connection["id"])
+        await update.message.reply_text("Telegram connected to DeerFlow.")
+        return True
+
+    async def _attach_connection_identity(self, inbound: InboundMessage) -> InboundMessage:
+        if self._connection_repo is None:
+            return inbound
+
+        connection = await self._connection_repo.find_connection_by_external_identity(
+            provider="telegram",
+            external_account_id=inbound.user_id,
+            workspace_id=inbound.chat_id,
+        )
+        if connection is None:
+            return inbound
+
+        inbound.connection_id = connection["id"]
+        inbound.owner_user_id = connection["owner_user_id"]
+        inbound.workspace_id = connection.get("workspace_id")
+        return inbound
+
+    def _get_bot_username(self, context) -> str | None:
+        bot = getattr(context, "bot", None)
+        username = getattr(bot, "username", None)
+        if not username and self._application is not None:
+            username = getattr(getattr(self._application, "bot", None), "username", None)
+        return str(username) if username else None
+
+    @staticmethod
+    def _strip_bot_username_from_leading_command(text: str, bot_username: str | None) -> str:
+        username = (bot_username or "").lstrip("@").lower()
+        if not username or not text.startswith("/"):
+            return text
+
+        parts = text.split(maxsplit=1)
+        command_token = parts[0]
+        if "@" not in command_token:
+            return text
+
+        command_name, addressed_username = command_token[1:].rsplit("@", 1)
+        if not command_name or addressed_username.lower() != username:
+            return text
+
+        normalized = f"/{command_name}"
+        if len(parts) > 1:
+            normalized = f"{normalized} {parts[1]}"
+        return normalized
+
     async def _cmd_start(self, update, context) -> None:
         """Handle /start command."""
         if not self._check_user(update.effective_user.id):
             return
+        args = getattr(context, "args", []) if context is not None else []
+        if args:
+            handled = await self._bind_connection_from_start_token(update, str(args[0]))
+            if handled:
+                return
         await update.message.reply_text("Welcome to DeerFlow! Send me a message to start a conversation.\nType /help for available commands.")
 
     async def _process_incoming_with_reply(self, chat_id: str, msg_id: int, inbound: InboundMessage) -> None:
@@ -243,7 +358,7 @@ class TelegramChannel(Channel):
         if not self._check_user(update.effective_user.id):
             return
 
-        text = update.message.text
+        text = self._strip_bot_username_from_leading_command(update.message.text.strip(), self._get_bot_username(context))
         chat_id = str(update.effective_chat.id)
         user_id = str(update.effective_user.id)
         msg_id = str(update.message.message_id)
@@ -267,6 +382,7 @@ class TelegramChannel(Channel):
             thread_ts=msg_id,
         )
         inbound.topic_id = topic_id
+        inbound = await self._attach_connection_identity(inbound)
 
         if self._main_loop and self._main_loop.is_running():
             fut = asyncio.run_coroutine_threadsafe(self._process_incoming_with_reply(chat_id, update.message.message_id, inbound), self._main_loop)
@@ -279,7 +395,7 @@ class TelegramChannel(Channel):
         if not self._check_user(update.effective_user.id):
             return
 
-        text = update.message.text.strip()
+        text = self._strip_bot_username_from_leading_command(update.message.text.strip(), self._get_bot_username(context))
         if not text:
             return
 
@@ -309,6 +425,7 @@ class TelegramChannel(Channel):
             thread_ts=msg_id,
         )
         inbound.topic_id = topic_id
+        inbound = await self._attach_connection_identity(inbound)
 
         if self._main_loop and self._main_loop.is_running():
             fut = asyncio.run_coroutine_threadsafe(self._process_incoming_with_reply(chat_id, update.message.message_id, inbound), self._main_loop)

backend/app/channels/wechat.py

@@ -22,6 +22,7 @@ from cryptography.hazmat.primitives import padding
 from cryptography.hazmat.primitives.ciphers import Cipher, algorithms, modes
 
 from app.channels.base import Channel
+from app.channels.commands import is_known_channel_command
 from app.channels.message_bus import InboundMessageType, MessageBus, OutboundMessage, ResolvedAttachment
 
 logger = logging.getLogger(__name__)
@@ -620,7 +621,7 @@ class WechatChannel(Channel):
             chat_id=chat_id,
             user_id=chat_id,
             text=text,
-            msg_type=InboundMessageType.COMMAND if text.startswith("/") else InboundMessageType.CHAT,
+            msg_type=InboundMessageType.COMMAND if is_known_channel_command(text) else InboundMessageType.CHAT,
             thread_ts=thread_ts,
             files=files,
             metadata={

backend/app/channels/wecom.py

@@ -8,6 +8,7 @@ from collections.abc import Awaitable, Callable
 from typing import Any, cast
 
 from app.channels.base import Channel
+from app.channels.commands import is_known_channel_command
 from app.channels.message_bus import (
     InboundMessageType,
     MessageBus,
@@ -270,7 +271,7 @@ class WeComChannel(Channel):
 
         user_id = (body.get("from") or {}).get("userid")
 
-        inbound_type = InboundMessageType.COMMAND if text.startswith("/") else InboundMessageType.CHAT
+        inbound_type = InboundMessageType.COMMAND if is_known_channel_command(text) else InboundMessageType.CHAT
         inbound = self._make_inbound(
             chat_id=user_id,  # keep user's conversation in memory
             user_id=user_id,

backend/app/gateway/app.py

@@ -6,6 +6,7 @@ from contextlib import asynccontextmanager
 from fastapi import FastAPI
 from fastapi.middleware.cors import CORSMiddleware
 
+from app.gateway.auth_disabled import warn_if_auth_disabled_enabled
 from app.gateway.auth_middleware import AuthMiddleware
 from app.gateway.config import get_gateway_config
 from app.gateway.csrf_middleware import CSRFMiddleware, get_configured_cors_origins
@@ -15,6 +16,7 @@ from app.gateway.routers import (
     artifacts,
     assistants_compat,
     auth,
+    channel_connections,
     channels,
     feedback,
     mcp,
@@ -172,6 +174,7 @@ async def lifespan(app: FastAPI) -> AsyncGenerator[None, None]:
         startup_config = get_app_config()
         apply_logging_level(startup_config.log_level)
         logger.info("Configuration loaded successfully")
+        warn_if_auth_disabled_enabled()
     except Exception as e:
         error_msg = f"Failed to load configuration during gateway startup: {e}"
         logger.exception(error_msg)
@@ -179,6 +182,25 @@ async def lifespan(app: FastAPI) -> AsyncGenerator[None, None]:
     config = get_gateway_config()
     logger.info(f"Starting API Gateway on {config.host}:{config.port}")
 
+    # Pre-warm tiktoken encoding cache so the first memory-injection request
+    # never blocks on the BPE data download (which hits an OpenAI/Azure URL
+    # that may be unreachable in restricted networks — see issue #3402).
+    try:
+        from deerflow.agents.memory.prompt import warm_tiktoken_cache
+
+        warmed = await asyncio.wait_for(
+            asyncio.to_thread(warm_tiktoken_cache),
+            timeout=5,
+        )
+        if warmed:
+            logger.info("tiktoken encoding cache warmed successfully")
+        else:
+            logger.warning("tiktoken encoding cache warm-up failed; token counting will use character-based fallback")
+    except TimeoutError:
+        logger.warning("tiktoken encoding cache warm-up timed out; token counting will use character-based fallback")
+    except Exception:
+        logger.warning("tiktoken warm-up skipped", exc_info=True)
+
     # Initialize LangGraph runtime components (StreamBridge, RunManager, checkpointer, store)
     async with langgraph_runtime(app, startup_config):
         logger.info("LangGraph runtime initialised")
@@ -357,6 +379,9 @@ This gateway provides runtime endpoints for agent runs plus custom endpoints for
     # Suggestions API is mounted at /api/threads/{thread_id}/suggestions
     app.include_router(suggestions.router)
 
+    # User-facing IM channel connection API is mounted at /api/channels
+    app.include_router(channel_connections.router)
+
     # Channels API is mounted at /api/channels
     app.include_router(channels.router)
 

backend/app/gateway/auth_disabled.py

@@ -0,0 +1,54 @@
+"""Shared helpers for local/E2E auth-disabled mode."""
+
+from __future__ import annotations
+
+import logging
+import os
+from types import SimpleNamespace
+
+AUTH_DISABLED_ENV_VAR = "DEER_FLOW_AUTH_DISABLED"
+AUTH_DISABLED_USER_ID = "e2e-user"
+AUTH_DISABLED_USER_EMAIL = "e2e@test.local"
+
+AUTH_SOURCE_SESSION = "session"
+AUTH_SOURCE_INTERNAL = "internal"
+AUTH_SOURCE_AUTH_DISABLED = "auth_disabled"
+
+_PRODUCTION_ENV_VARS: tuple[str, ...] = ("DEER_FLOW_ENV", "ENVIRONMENT")
+_PRODUCTION_ENV_VALUES: frozenset[str] = frozenset({"prod", "production"})
+
+logger = logging.getLogger(__name__)
+
+
+def is_explicit_production_environment() -> bool:
+    return any(os.environ.get(name, "").strip().lower() in _PRODUCTION_ENV_VALUES for name in _PRODUCTION_ENV_VARS)
+
+
+def is_auth_disabled_requested() -> bool:
+    return os.environ.get(AUTH_DISABLED_ENV_VAR) == "1"
+
+
+def is_auth_disabled() -> bool:
+    return is_auth_disabled_requested() and not is_explicit_production_environment()
+
+
+def warn_if_auth_disabled_enabled() -> None:
+    if not is_auth_disabled():
+        return
+
+    logger.warning(
+        "%s=1 is active: authentication is bypassed and anonymous requests run as synthetic admin user %r. Do not enable this in shared or production deployments.",
+        AUTH_DISABLED_ENV_VAR,
+        AUTH_DISABLED_USER_ID,
+    )
+
+
+def get_auth_disabled_user():
+    return SimpleNamespace(
+        id=AUTH_DISABLED_USER_ID,
+        email=AUTH_DISABLED_USER_EMAIL,
+        password_hash=None,
+        system_role="admin",
+        needs_setup=False,
+        token_version=0,
+    )

backend/app/gateway/auth_middleware.py

@@ -17,6 +17,13 @@ from starlette.responses import JSONResponse
 from starlette.types import ASGIApp
 
 from app.gateway.auth.errors import AuthErrorCode, AuthErrorResponse
+from app.gateway.auth_disabled import (
+    AUTH_SOURCE_AUTH_DISABLED,
+    AUTH_SOURCE_INTERNAL,
+    AUTH_SOURCE_SESSION,
+    get_auth_disabled_user,
+    is_auth_disabled,
+)
 from app.gateway.authz import _ALL_PERMISSIONS, AuthContext
 from app.gateway.internal_auth import INTERNAL_AUTH_HEADER_NAME, get_internal_user, is_valid_internal_auth_token
 from deerflow.runtime.user_context import reset_current_user, set_current_user
@@ -80,8 +87,38 @@ class AuthMiddleware(BaseHTTPMiddleware):
         if is_valid_internal_auth_token(request.headers.get(INTERNAL_AUTH_HEADER_NAME)):
             internal_user = get_internal_user()
 
+        auth_source = AUTH_SOURCE_SESSION
+        access_token = request.cookies.get("access_token")
+
         # Non-public path: require session cookie
-        if internal_user is None and not request.cookies.get("access_token"):
+        if internal_user is not None:
+            user = internal_user
+            auth_source = AUTH_SOURCE_INTERNAL
+        elif access_token:
+            # Strict JWT validation: reject junk/expired tokens with 401
+            # right here instead of silently passing through. This closes
+            # the "junk cookie bypass" gap (AUTH_TEST_PLAN test 7.5.8):
+            # without this, non-isolation routes like /api/models would
+            # accept any cookie-shaped string as authentication.
+            #
+            # We call the *strict* resolver so that fine-grained error
+            # codes (token_expired, token_invalid, user_not_found, …)
+            # propagate from AuthErrorCode, not get flattened into one
+            # generic code. BaseHTTPMiddleware doesn't let HTTPException
+            # bubble up, so we catch and render it as JSONResponse here.
+            from app.gateway.deps import get_current_user_from_request
+
+            try:
+                user = await get_current_user_from_request(request)
+            except HTTPException as exc:
+                if not is_auth_disabled():
+                    return JSONResponse(status_code=exc.status_code, content={"detail": exc.detail})
+                user = get_auth_disabled_user()
+                auth_source = AUTH_SOURCE_AUTH_DISABLED
+        elif is_auth_disabled():
+            user = get_auth_disabled_user()
+            auth_source = AUTH_SOURCE_AUTH_DISABLED
+        else:
             return JSONResponse(
                 status_code=401,
                 content={
@@ -92,32 +129,12 @@ class AuthMiddleware(BaseHTTPMiddleware):
                 },
             )
 
-        # Strict JWT validation: reject junk/expired tokens with 401
-        # right here instead of silently passing through. This closes
-        # the "junk cookie bypass" gap (AUTH_TEST_PLAN test 7.5.8):
-        # without this, non-isolation routes like /api/models would
-        # accept any cookie-shaped string as authentication.
-        #
-        # We call the *strict* resolver so that fine-grained error
-        # codes (token_expired, token_invalid, user_not_found, …)
-        # propagate from AuthErrorCode, not get flattened into one
-        # generic code. BaseHTTPMiddleware doesn't let HTTPException
-        # bubble up, so we catch and render it as JSONResponse here.
-        from app.gateway.deps import get_current_user_from_request
-
-        if internal_user is not None:
-            user = internal_user
-        else:
-            try:
-                user = await get_current_user_from_request(request)
-            except HTTPException as exc:
-                return JSONResponse(status_code=exc.status_code, content={"detail": exc.detail})
-
         # Stamp both request.state.user (for the contextvar pattern)
         # and request.state.auth (so @require_permission's "auth is
         # None" branch short-circuits instead of running the entire
         # JWT-decode + DB-lookup pipeline a second time per request).
         request.state.user = user
+        request.state.auth_source = auth_source
         request.state.auth = AuthContext(user=user, permissions=_ALL_PERMISSIONS)
         token = set_current_user(user)
         try:

backend/app/gateway/csrf_middleware.py

@@ -14,6 +14,8 @@ from starlette.middleware.base import BaseHTTPMiddleware
 from starlette.responses import JSONResponse
 from starlette.types import ASGIApp
 
+from app.gateway.auth_disabled import is_auth_disabled
+
 CSRF_COOKIE_NAME = "csrf_token"
 CSRF_HEADER_NAME = "X-CSRF-Token"
 CSRF_TOKEN_LENGTH = 64  # bytes
@@ -38,6 +40,9 @@ def should_check_csrf(request: Request) -> bool:
     if request.method not in ("POST", "PUT", "DELETE", "PATCH"):
         return False
 
+    if is_auth_disabled():
+        return False
+
     path = request.url.path.rstrip("/")
     # Exempt /api/v1/auth/me endpoint
     if path == "/api/v1/auth/me":

backend/app/gateway/deps.py

@@ -331,6 +331,17 @@ async def get_current_user_from_request(request: Request):
 
     Raises HTTPException 401 if not authenticated.
     """
+    state = getattr(request, "state", None)
+    state_user = getattr(state, "user", None)
+    from app.gateway.auth_disabled import AUTH_SOURCE_AUTH_DISABLED, AUTH_SOURCE_INTERNAL, AUTH_SOURCE_SESSION
+
+    if state_user is not None and getattr(state, "auth_source", None) in {
+        AUTH_SOURCE_SESSION,
+        AUTH_SOURCE_AUTH_DISABLED,
+        AUTH_SOURCE_INTERNAL,
+    }:
+        return state_user
+
     from app.gateway.auth import decode_token
     from app.gateway.auth.errors import AuthErrorCode, AuthErrorResponse, TokenError, token_error_to_code
 

backend/app/gateway/langgraph_auth.py

@@ -20,6 +20,7 @@ from langgraph_sdk import Auth
 
 from app.gateway.auth.errors import TokenError
 from app.gateway.auth.jwt import decode_token
+from app.gateway.auth_disabled import AUTH_DISABLED_USER_ID, is_auth_disabled
 from app.gateway.deps import get_local_provider
 
 auth = Auth()
@@ -38,6 +39,9 @@ def _check_csrf(request) -> None:
     if method.upper() not in _CSRF_METHODS:
         return
 
+    if is_auth_disabled():
+        return
+
     cookie_token = request.cookies.get("csrf_token")
     header_token = request.headers.get("x-csrf-token")
 
@@ -66,6 +70,9 @@ async def authenticate(request):
     # are rejected early, even if the cookie carries a valid JWT.
     _check_csrf(request)
 
+    if is_auth_disabled():
+        return AUTH_DISABLED_USER_ID
+
     token = request.cookies.get("access_token")
     if not token:
         raise Auth.exceptions.HTTPException(

backend/app/gateway/routers/agents.py

@@ -1,5 +1,6 @@
 """CRUD API for custom agents."""
 
+import asyncio
 import logging
 import re
 import shutil
@@ -213,48 +214,61 @@ async def create_agent_endpoint(request: AgentCreateRequest) -> AgentResponse:
     user_id = get_effective_user_id()
     paths = get_paths()
 
-    agent_dir = paths.user_agent_dir(user_id, normalized_name)
-    legacy_dir = paths.agent_dir(normalized_name)
+    def _create_agent() -> AgentResponse | None:
+        # Worker thread: base-dir resolution, existence checks, directory/file
+        # creation, read-back, and failure cleanup are all blocking filesystem
+        # IO that must stay off the event loop.
+        agent_dir = paths.user_agent_dir(user_id, normalized_name)
+        legacy_dir = paths.agent_dir(normalized_name)
 
-    if agent_dir.exists() or legacy_dir.exists():
-        raise HTTPException(status_code=409, detail=f"Agent '{normalized_name}' already exists")
+        if legacy_dir.exists():
+            return None  # signals 409 to the caller
+
+        try:
+            try:
+                agent_dir.mkdir(parents=True, exist_ok=False)
+            except FileExistsError:
+                return None  # signals 409 to the caller
+            # Write config.yaml
+            config_data: dict = {"name": normalized_name}
+            if request.description:
+                config_data["description"] = request.description
+            if request.model is not None:
+                config_data["model"] = request.model
+            if request.tool_groups is not None:
+                config_data["tool_groups"] = request.tool_groups
+            if request.skills is not None:
+                config_data["skills"] = request.skills
+
+            config_file = agent_dir / "config.yaml"
+            with open(config_file, "w", encoding="utf-8") as f:
+                yaml.dump(config_data, f, default_flow_style=False, allow_unicode=True)
+
+            # Write SOUL.md
+            soul_file = agent_dir / "SOUL.md"
+            soul_file.write_text(request.soul, encoding="utf-8")
+
+            logger.info(f"Created agent '{normalized_name}' at {agent_dir}")
+
+            agent_cfg = load_agent_config(normalized_name, user_id=user_id)
+            return _agent_config_to_response(agent_cfg, include_soul=True, user_id=user_id)
+        except Exception:
+            # Clean up partial state on failure before surfacing the error.
+            if agent_dir.exists():
+                shutil.rmtree(agent_dir)
+            raise
 
     try:
-        agent_dir.mkdir(parents=True, exist_ok=True)
-
-        # Write config.yaml
-        config_data: dict = {"name": normalized_name}
-        if request.description:
-            config_data["description"] = request.description
-        if request.model is not None:
-            config_data["model"] = request.model
-        if request.tool_groups is not None:
-            config_data["tool_groups"] = request.tool_groups
-        if request.skills is not None:
-            config_data["skills"] = request.skills
-
-        config_file = agent_dir / "config.yaml"
-        with open(config_file, "w", encoding="utf-8") as f:
-            yaml.dump(config_data, f, default_flow_style=False, allow_unicode=True)
-
-        # Write SOUL.md
-        soul_file = agent_dir / "SOUL.md"
-        soul_file.write_text(request.soul, encoding="utf-8")
-
-        logger.info(f"Created agent '{normalized_name}' at {agent_dir}")
-
-        agent_cfg = load_agent_config(normalized_name, user_id=user_id)
-        return _agent_config_to_response(agent_cfg, include_soul=True, user_id=user_id)
-
-    except HTTPException:
-        raise
+        response = await asyncio.to_thread(_create_agent)
     except Exception as e:
-        # Clean up on failure
-        if agent_dir.exists():
-            shutil.rmtree(agent_dir)
         logger.error(f"Failed to create agent '{request.name}': {e}", exc_info=True)
         raise HTTPException(status_code=500, detail=f"Failed to create agent: {str(e)}")
 
+    if response is None:
+        raise HTTPException(status_code=409, detail=f"Agent '{normalized_name}' already exists")
+
+    return response
+
 
 @router.put(
     "/agents/{name}",
@@ -428,19 +442,30 @@ async def delete_agent(name: str) -> None:
     name = _normalize_agent_name(name)
     user_id = get_effective_user_id()
     paths = get_paths()
-    agent_dir = paths.user_agent_dir(user_id, name)
 
-    if not agent_dir.exists():
-        if paths.agent_dir(name).exists():
-            raise HTTPException(
-                status_code=409,
-                detail=(f"Agent '{name}' only exists in the legacy shared layout and is not scoped to a user. Run scripts/migrate_user_isolation.py to move legacy agents into the per-user layout before deleting."),
-            )
-        raise HTTPException(status_code=404, detail=f"Agent '{name}' not found")
+    def _remove_agent_dir() -> tuple[str, str]:
+        # Runs in a worker thread: resolving the base dir, probing the directory
+        # (`exists`), and removing it (`rmtree`) are all blocking filesystem IO
+        # that must stay off the event loop.
+        agent_dir = paths.user_agent_dir(user_id, name)
+        if not agent_dir.exists():
+            outcome = "legacy" if paths.agent_dir(name).exists() else "missing"
+            return outcome, str(agent_dir)
+        shutil.rmtree(agent_dir)
+        return "deleted", str(agent_dir)
 
     try:
-        shutil.rmtree(agent_dir)
-        logger.info(f"Deleted agent '{name}' from {agent_dir}")
+        outcome, agent_dir = await asyncio.to_thread(_remove_agent_dir)
     except Exception as e:
         logger.error(f"Failed to delete agent '{name}': {e}", exc_info=True)
         raise HTTPException(status_code=500, detail=f"Failed to delete agent: {str(e)}")
+
+    if outcome == "legacy":
+        raise HTTPException(
+            status_code=409,
+            detail=(f"Agent '{name}' only exists in the legacy shared layout and is not scoped to a user. Run scripts/migrate_user_isolation.py to move legacy agents into the per-user layout before deleting."),
+        )
+    if outcome == "missing":
+        raise HTTPException(status_code=404, detail=f"Agent '{name}' not found")
+
+    logger.info(f"Deleted agent '{name}' from {agent_dir}")

backend/app/gateway/routers/auth.py

@@ -341,9 +341,19 @@ async def change_password(request: Request, response: Response, body: ChangePass
     - Re-issues session cookie with new token_version
     """
     from app.gateway.auth.password import hash_password_async, verify_password_async
+    from app.gateway.auth_disabled import AUTH_SOURCE_AUTH_DISABLED
 
     user = await get_current_user_from_request(request)
 
+    if getattr(request.state, "auth_source", None) == AUTH_SOURCE_AUTH_DISABLED:
+        raise HTTPException(
+            status_code=status.HTTP_400_BAD_REQUEST,
+            detail=AuthErrorResponse(
+                code=AuthErrorCode.INVALID_CREDENTIALS,
+                message="Password changes are not available when DEER_FLOW_AUTH_DISABLED=1.",
+            ).model_dump(),
+        )
+
     if user.password_hash is None:
         raise HTTPException(status_code=status.HTTP_400_BAD_REQUEST, detail=AuthErrorResponse(code=AuthErrorCode.INVALID_CREDENTIALS, message="OAuth users cannot change password").model_dump())
 

backend/app/gateway/routers/channel_connections.py

@@ -0,0 +1,294 @@
+"""Browser-facing APIs for user-owned IM channel bindings."""
+
+from __future__ import annotations
+
+import secrets
+from datetime import UTC, datetime, timedelta
+from typing import Any
+
+from fastapi import APIRouter, HTTPException, Request, Response
+from pydantic import BaseModel, Field
+
+from deerflow.config.channel_connections_config import ChannelConnectionsConfig
+from deerflow.persistence.channel_connections import ChannelConnectionRepository
+from deerflow.persistence.engine import get_session_factory
+
+router = APIRouter(prefix="/api/channels", tags=["channel-connections"])
+
+_STATE_TTL_SECONDS = 600
+
+
+class ChannelProviderResponse(BaseModel):
+    provider: str
+    display_name: str
+    enabled: bool
+    configured: bool
+    connectable: bool
+    unavailable_reason: str | None = None
+    auth_mode: str
+    connection_status: str
+
+
+class ChannelProvidersResponse(BaseModel):
+    enabled: bool
+    providers: list[ChannelProviderResponse]
+
+
+class ChannelConnectionResponse(BaseModel):
+    id: str
+    provider: str
+    status: str
+    external_account_id: str | None = None
+    external_account_name: str | None = None
+    workspace_id: str | None = None
+    workspace_name: str | None = None
+    scopes: list[str] = Field(default_factory=list)
+    metadata: dict[str, Any] = Field(default_factory=dict)
+
+
+class ChannelConnectionsResponse(BaseModel):
+    connections: list[ChannelConnectionResponse]
+
+
+class ChannelConnectResponse(BaseModel):
+    provider: str
+    mode: str
+    url: str | None = None
+    code: str
+    instruction: str
+    expires_in: int
+
+
+_PROVIDER_META: dict[str, dict[str, str]] = {
+    "telegram": {"display_name": "Telegram", "auth_mode": "deep_link"},
+    "slack": {"display_name": "Slack", "auth_mode": "binding_code"},
+    "discord": {"display_name": "Discord", "auth_mode": "binding_code"},
+}
+
+_RUNTIME_REQUIREMENTS: dict[str, tuple[str, ...]] = {
+    "telegram": ("bot_token",),
+    "slack": ("bot_token", "app_token"),
+    "discord": ("bot_token",),
+}
+
+
+def _get_user_id(request: Request) -> str:
+    user = getattr(request.state, "user", None)
+    if user is None:
+        raise HTTPException(status_code=401, detail="Authentication required")
+    return str(user.id)
+
+
+def _get_app_config():
+    from deerflow.config.app_config import get_app_config
+
+    return get_app_config()
+
+
+def _get_channel_connections_config(request: Request) -> ChannelConnectionsConfig:
+    config = getattr(request.app.state, "channel_connections_config", None)
+    if isinstance(config, ChannelConnectionsConfig):
+        return config
+    return _get_app_config().channel_connections
+
+
+def _get_channels_config(request: Request) -> dict[str, Any]:
+    state_config = getattr(request.app.state, "channels_config", None)
+    if isinstance(state_config, dict):
+        return state_config
+
+    app_config = _get_app_config()
+    extra = app_config.model_extra or {}
+    channels_config = extra.get("channels")
+    return dict(channels_config) if isinstance(channels_config, dict) else {}
+
+
+def _get_repository(request: Request, config: ChannelConnectionsConfig) -> ChannelConnectionRepository:
+    repo = getattr(request.app.state, "channel_connection_repo", None)
+    if isinstance(repo, ChannelConnectionRepository):
+        return repo
+
+    sf = get_session_factory()
+    if sf is None:
+        raise HTTPException(status_code=503, detail="Channel connection persistence is not available")
+
+    repo = ChannelConnectionRepository(sf)
+    request.app.state.channel_connection_repo = repo
+    return repo
+
+
+def _provider_config(config: ChannelConnectionsConfig, provider: str):
+    provider_config = getattr(config, provider, None)
+    if provider_config is None:
+        raise HTTPException(status_code=404, detail="Unknown channel provider")
+    return provider_config
+
+
+def _runtime_channel_configured(provider: str, channels_config: dict[str, Any]) -> bool:
+    runtime_config = channels_config.get(provider)
+    if not isinstance(runtime_config, dict) or not runtime_config.get("enabled", False):
+        return False
+    return all(str(runtime_config.get(key) or "").strip() for key in _RUNTIME_REQUIREMENTS[provider])
+
+
+def _runtime_unavailable_reason(provider: str) -> str:
+    keys = " and ".join(f"channels.{provider}.{key}" for key in _RUNTIME_REQUIREMENTS[provider])
+    return f"Enable and configure channels.{provider} with {keys}."
+
+
+def _provider_unavailable_reason(
+    config: ChannelConnectionsConfig,
+    channels_config: dict[str, Any],
+    provider: str,
+) -> str | None:
+    provider_config = _provider_config(config, provider)
+    if not provider_config.enabled:
+        return None
+    if not provider_config.configured:
+        if provider == "telegram":
+            return "Configure channel_connections.telegram.bot_username for Telegram deep links."
+        return f"Configure channel_connections.{provider}."
+    if not _runtime_channel_configured(provider, channels_config):
+        return _runtime_unavailable_reason(provider)
+    return None
+
+
+def _provider_status(
+    config: ChannelConnectionsConfig,
+    channels_config: dict[str, Any],
+    provider: str,
+) -> tuple[dict[str, bool], str | None]:
+    declared = config.provider_status(provider)
+    unavailable_reason = _provider_unavailable_reason(config, channels_config, provider)
+    configured = declared["configured"] and _runtime_channel_configured(provider, channels_config)
+    return {"enabled": declared["enabled"], "configured": configured}, unavailable_reason
+
+
+def _new_binding_code() -> str:
+    return secrets.token_hex(4)
+
+
+async def _create_state(
+    repo: ChannelConnectionRepository,
+    *,
+    owner_user_id: str,
+    provider: str,
+) -> str:
+    state = _new_binding_code()
+    await repo.create_oauth_state(
+        owner_user_id=owner_user_id,
+        provider=provider,
+        state=state,
+        expires_at=datetime.now(UTC) + timedelta(seconds=_STATE_TTL_SECONDS),
+    )
+    return state
+
+
+def _connect_instruction(provider: str, code: str) -> str:
+    if provider == "telegram":
+        return f"Send /start {code} to the DeerFlow Telegram bot."
+    if provider == "slack":
+        return f"Send /connect {code} to the DeerFlow Slack bot."
+    if provider == "discord":
+        return f"Send /connect {code} to the DeerFlow Discord bot."
+    raise HTTPException(status_code=404, detail="Unknown channel provider")
+
+
+def _connect_url(config: ChannelConnectionsConfig, provider: str, code: str) -> str | None:
+    if provider == "telegram":
+        provider_config = _provider_config(config, provider)
+        return f"https://t.me/{provider_config.bot_username}?start={code}"
+    if provider in {"slack", "discord"}:
+        return None
+    raise HTTPException(status_code=404, detail="Unknown channel provider")
+
+
+@router.get("/providers", response_model=ChannelProvidersResponse)
+async def get_channel_providers(request: Request) -> ChannelProvidersResponse:
+    config = _get_channel_connections_config(request)
+    channels_config = _get_channels_config(request)
+    repo = None
+    if config.enabled:
+        try:
+            repo = _get_repository(request, config)
+        except HTTPException as exc:
+            if exc.status_code != 503:
+                raise
+    owner_user_id = _get_user_id(request)
+    connections = await repo.list_connections(owner_user_id) if repo is not None else []
+    by_provider = {item["provider"]: item for item in connections}
+
+    providers: list[ChannelProviderResponse] = []
+    for provider, meta in _PROVIDER_META.items():
+        status, unavailable_reason = _provider_status(config, channels_config, provider)
+        connection = by_provider.get(provider)
+        providers.append(
+            ChannelProviderResponse(
+                provider=provider,
+                display_name=meta["display_name"],
+                enabled=status["enabled"],
+                configured=status["configured"],
+                connectable=status["enabled"] and status["configured"] and unavailable_reason is None,
+                unavailable_reason=unavailable_reason,
+                auth_mode=meta["auth_mode"],
+                connection_status=connection["status"] if connection else "not_connected",
+            )
+        )
+    return ChannelProvidersResponse(enabled=config.enabled, providers=providers)
+
+
+@router.get("/connections", response_model=ChannelConnectionsResponse)
+async def get_channel_connections(request: Request) -> ChannelConnectionsResponse:
+    config = _get_channel_connections_config(request)
+    if not config.enabled:
+        return ChannelConnectionsResponse(connections=[])
+    repo = _get_repository(request, config)
+    rows = await repo.list_connections(_get_user_id(request))
+    return ChannelConnectionsResponse(connections=[ChannelConnectionResponse(**row) for row in rows])
+
+
+@router.delete("/connections/{connection_id}", status_code=204)
+async def disconnect_channel_connection(connection_id: str, request: Request) -> Response:
+    config = _get_channel_connections_config(request)
+    if not config.enabled:
+        raise HTTPException(status_code=400, detail="Channel connections are disabled")
+
+    repo = _get_repository(request, config)
+    disconnected = await repo.disconnect_connection(
+        connection_id=connection_id,
+        owner_user_id=_get_user_id(request),
+    )
+    if not disconnected:
+        raise HTTPException(status_code=404, detail="Channel connection not found")
+    return Response(status_code=204)
+
+
+@router.post("/{provider}/connect", response_model=ChannelConnectResponse)
+async def connect_channel_provider(provider: str, request: Request) -> ChannelConnectResponse:
+    config = _get_channel_connections_config(request)
+    channels_config = _get_channels_config(request)
+    if not config.enabled:
+        raise HTTPException(status_code=400, detail="Channel connections are disabled")
+
+    status, unavailable_reason = _provider_status(config, channels_config, provider)
+    if not status["enabled"]:
+        raise HTTPException(status_code=400, detail="Channel provider is not enabled")
+    if unavailable_reason:
+        raise HTTPException(status_code=400, detail=unavailable_reason)
+    if not status["configured"]:
+        raise HTTPException(status_code=400, detail="Channel provider is not configured")
+
+    repo = _get_repository(request, config)
+    code = await _create_state(
+        repo,
+        owner_user_id=_get_user_id(request),
+        provider=provider,
+    )
+    return ChannelConnectResponse(
+        provider=provider,
+        mode=_PROVIDER_META[provider]["auth_mode"],
+        url=_connect_url(config, provider, code),
+        code=code,
+        instruction=_connect_instruction(provider, code),
+        expires_in=_STATE_TTL_SECONDS,
+    )

backend/app/gateway/routers/mcp.py

@@ -1,9 +1,10 @@
 import json
 import logging
+import os
 from pathlib import Path
 from typing import Literal
 
-from fastapi import APIRouter, HTTPException
+from fastapi import APIRouter, HTTPException, Request, status
 from pydantic import BaseModel, Field
 
 from deerflow.config.extensions_config import ExtensionsConfig, get_extensions_config, reload_extensions_config
@@ -12,6 +13,11 @@ logger = logging.getLogger(__name__)
 router = APIRouter(prefix="/api", tags=["mcp"])
 
 
+_MCP_STDIO_COMMAND_ALLOWLIST_ENV = "DEER_FLOW_MCP_STDIO_COMMAND_ALLOWLIST"
+_DEFAULT_MCP_STDIO_COMMAND_ALLOWLIST = frozenset({"npx", "uvx"})
+_SHELL_METACHARS = frozenset(";|&`$<>\n\r")
+
+
 class McpOAuthConfigResponse(BaseModel):
     """OAuth configuration for an MCP server."""
 
@@ -66,6 +72,78 @@ class McpConfigUpdateRequest(BaseModel):
 _MASKED_VALUE = "***"
 
 
+async def _require_admin_user(request: Request) -> None:
+    """Require the authenticated caller to be an admin user.
+
+    ``AuthMiddleware`` normally stamps ``request.state.user`` before the
+    request reaches this router. Falling back to the strict dependency keeps
+    this route safe even in tests or alternative ASGI compositions that mount
+    the router without the global middleware.
+    """
+    user = getattr(request.state, "user", None)
+    if user is None:
+        from app.gateway.deps import get_current_user_from_request
+
+        user = await get_current_user_from_request(request)
+
+    if getattr(user, "system_role", None) != "admin":
+        raise HTTPException(
+            status_code=status.HTTP_403_FORBIDDEN,
+            detail="Admin privileges required to manage MCP configuration.",
+        )
+
+
+def _allowed_stdio_commands() -> set[str]:
+    """Return executable names allowed for API-managed stdio MCP servers."""
+    raw = os.environ.get(_MCP_STDIO_COMMAND_ALLOWLIST_ENV)
+    base = set(_DEFAULT_MCP_STDIO_COMMAND_ALLOWLIST)
+    if raw is None:
+        return base
+    extra = {item.strip() for item in raw.split(",") if item.strip()}
+    return base | extra
+
+
+def _stdio_command_name(command: str | None, *, server_name: str) -> str:
+    """Normalize and validate a stdio command field from the API boundary."""
+    if command is None or not command.strip():
+        raise HTTPException(
+            status_code=status.HTTP_400_BAD_REQUEST,
+            detail=f"MCP server '{server_name}' with stdio transport requires a command.",
+        )
+
+    stripped = command.strip()
+    has_path_separator = "/" in stripped or "\\" in stripped
+    if stripped != command or has_path_separator or any(ch.isspace() for ch in stripped) or any(ch in stripped for ch in _SHELL_METACHARS):
+        raise HTTPException(
+            status_code=status.HTTP_400_BAD_REQUEST,
+            detail=(f"MCP server '{server_name}' command must be a single executable name; put parameters in args instead."),
+        )
+
+    return stripped
+
+
+def _validate_mcp_update_request(request: McpConfigUpdateRequest) -> None:
+    """Validate API-submitted MCP config before it is persisted.
+
+    Local config files can still express arbitrary advanced setups, but the
+    HTTP API is an untrusted boundary. Restricting stdio commands here reduces
+    the blast radius of a compromised authenticated browser session.
+    """
+    allowed_commands = _allowed_stdio_commands()
+    for name, server in request.mcp_servers.items():
+        transport_type = (server.type or "stdio").lower()
+        if transport_type != "stdio":
+            continue
+
+        command_name = _stdio_command_name(server.command, server_name=name)
+        if command_name not in allowed_commands:
+            allowed = ", ".join(sorted(allowed_commands)) or "<none>"
+            raise HTTPException(
+                status_code=status.HTTP_400_BAD_REQUEST,
+                detail=(f"MCP server '{name}' uses disallowed stdio command '{command_name}'. Allowed commands: {allowed}. Configure {_MCP_STDIO_COMMAND_ALLOWLIST_ENV} to extend this list."),
+            )
+
+
 def _mask_server_config(server: McpServerConfigResponse) -> McpServerConfigResponse:
     """Return a copy of server config with sensitive fields masked.
 
@@ -162,7 +240,7 @@ def _merge_preserving_secrets(
     summary="Get MCP Configuration",
     description="Retrieve the current Model Context Protocol (MCP) server configurations.",
 )
-async def get_mcp_configuration() -> McpConfigResponse:
+async def get_mcp_configuration(request: Request) -> McpConfigResponse:
     """Get the current MCP configuration.
 
     Returns:
@@ -183,6 +261,8 @@ async def get_mcp_configuration() -> McpConfigResponse:
         }
         ```
     """
+    await _require_admin_user(request)
+
     config = get_extensions_config()
 
     servers = {name: _mask_server_config(McpServerConfigResponse(**server.model_dump())) for name, server in config.mcp_servers.items()}
@@ -195,7 +275,7 @@ async def get_mcp_configuration() -> McpConfigResponse:
     summary="Update MCP Configuration",
     description="Update Model Context Protocol (MCP) server configurations and save to file.",
 )
-async def update_mcp_configuration(request: McpConfigUpdateRequest) -> McpConfigResponse:
+async def update_mcp_configuration(request: Request, body: McpConfigUpdateRequest) -> McpConfigResponse:
     """Update the MCP configuration.
 
     This will:
@@ -228,6 +308,9 @@ async def update_mcp_configuration(request: McpConfigUpdateRequest) -> McpConfig
         ```
     """
     try:
+        await _require_admin_user(request)
+        _validate_mcp_update_request(body)
+
         # Get the current config path (or determine where to save it)
         config_path = ExtensionsConfig.resolve_config_path()
 
@@ -255,7 +338,7 @@ async def update_mcp_configuration(request: McpConfigUpdateRequest) -> McpConfig
 
         # Merge incoming server configs with raw on-disk secrets
         merged_servers: dict[str, McpServerConfigResponse] = {}
-        for name, incoming in request.mcp_servers.items():
+        for name, incoming in body.mcp_servers.items():
             raw_server = raw_servers.get(name)
             if raw_server is not None:
                 merged_servers[name] = _merge_preserving_secrets(
@@ -283,6 +366,8 @@ async def update_mcp_configuration(request: McpConfigUpdateRequest) -> McpConfig
         servers = {name: _mask_server_config(McpServerConfigResponse(**server.model_dump())) for name, server in reloaded_config.mcp_servers.items()}
         return McpConfigResponse(mcp_servers=servers)
 
+    except HTTPException:
+        raise
     except Exception as e:
         logger.error(f"Failed to update MCP configuration: {e}", exc_info=True)
         raise HTTPException(status_code=500, detail=f"Failed to update MCP configuration: {str(e)}")

backend/app/gateway/routers/suggestions.py

@@ -1,5 +1,6 @@
 import json
 import logging
+import re
 
 from fastapi import APIRouter, Depends, Request
 from langchain_core.messages import HumanMessage, SystemMessage
@@ -30,6 +31,31 @@ class SuggestionsResponse(BaseModel):
     suggestions: list[str] = Field(default_factory=list, description="Suggested follow-up questions")
 
 
+# Matches a complete <think>...</think> block (case-insensitive, spans newlines).
+_THINK_BLOCK_RE = re.compile(r"<think\b[^>]*>.*?</think\s*>", re.IGNORECASE | re.DOTALL)
+# Matches a dangling, unclosed <think> (model truncated at max_tokens mid-thought).
+_OPEN_THINK_RE = re.compile(r"<think\b[^>]*>", re.IGNORECASE)
+
+
+def _strip_think_blocks(text: str) -> str:
+    """Remove reasoning-model ``<think>...</think>`` blocks from the response.
+
+    Reasoning models such as MiniMax-M3 inline their chain-of-thought into the
+    message ``content`` wrapped in ``<think>...</think>`` (``reasoning_split``
+    defaults to false), rather than exposing a separate ``reasoning_content``
+    field. The thinking text frequently contains ``[`` / ``]`` characters, which
+    corrupted the downstream ``find('[')`` / ``rfind(']')`` JSON extraction and
+    produced empty suggestions. We strip the reasoning before parsing so only
+    the actual answer remains.
+    """
+    text = _THINK_BLOCK_RE.sub("", text)
+    # Drop any unclosed <think> (and everything after it) left by truncation.
+    open_match = _OPEN_THINK_RE.search(text)
+    if open_match:
+        text = text[: open_match.start()]
+    return text.strip()
+
+
 def _strip_markdown_code_fence(text: str) -> str:
     stripped = text.strip()
     if not stripped.startswith("```"):
@@ -41,7 +67,8 @@ def _strip_markdown_code_fence(text: str) -> str:
 
 
 def _parse_json_string_list(text: str) -> list[str] | None:
-    candidate = _strip_markdown_code_fence(text)
+    candidate = _strip_think_blocks(text)
+    candidate = _strip_markdown_code_fence(candidate)
     start = candidate.find("[")
     end = candidate.rfind("]")
     if start == -1 or end == -1 or end <= start:

backend/app/gateway/routers/threads.py

@@ -17,7 +17,7 @@ import uuid
 from typing import Any
 
 from fastapi import APIRouter, HTTPException, Request
-from langgraph.checkpoint.base import empty_checkpoint
+from langgraph.checkpoint.base import empty_checkpoint, uuid6
 from pydantic import BaseModel, Field, field_validator
 
 from app.gateway.authz import require_permission
@@ -536,9 +536,21 @@ async def update_thread_state(thread_id: str, body: ThreadStateUpdateRequest, re
         metadata["step"] = metadata.get("step", 0) + 1
         metadata["writes"] = {body.as_node: body.values}
 
+    # Assign a new checkpoint ID so aput performs an INSERT rather than an
+    # in-place REPLACE of the existing row.  Use uuid6 (time-ordered) rather
+    # than uuid4 (random) so the new ID is always lexicographically greater
+    # than the previous one — LangGraph's checkpointers determine the "latest"
+    # checkpoint by max(checkpoint_ids) string order, matching the uuid6 epoch.
+    checkpoint["id"] = str(uuid6())
+
     # aput requires checkpoint_ns in the config — use the same config used for the
-    # read (which always includes checkpoint_ns="").  Do NOT include checkpoint_id
-    # so that aput generates a fresh checkpoint ID for the new snapshot.
+    # read (which always includes checkpoint_ns=""). The fresh checkpoint ID is
+    # assigned above via checkpoint["id"]; keep checkpoint_id out of the config so
+    # the write is keyed by the new checkpoint payload rather than the prior read.
+    # All supported savers (InMemorySaver, AsyncSqliteSaver, AsyncPostgresSaver)
+    # persist and echo back checkpoint["id"] verbatim — none mint their own — so
+    # the new_config below carries the uuid6 we assigned here. (Regression-locked
+    # by test_update_thread_state_inserts_new_checkpoint_each_call.)
     write_config: dict[str, Any] = {
         "configurable": {
             "thread_id": thread_id,
@@ -557,7 +569,7 @@ async def update_thread_state(thread_id: str, body: ThreadStateUpdateRequest, re
 
     # Sync title changes through the ThreadMetaStore abstraction so /threads/search
     # reflects them immediately in both sqlite and memory backends.
-    if body.values and "title" in body.values:
+    if thread_store and body.values and "title" in body.values:
         new_title = body.values["title"]
         if new_title:  # Skip empty strings and None
             try:

backend/docs/API.md

@@ -228,10 +228,13 @@ Get current MCP server configurations.
 GET /api/mcp/config
 ```
 
+Requires an authenticated admin session. Sensitive env/header/OAuth secret
+values are masked in the response.
+
 **Response:**
 ```json
 {
-  "mcpServers": {
+  "mcp_servers": {
     "github": {
       "enabled": true,
       "type": "stdio",
@@ -255,10 +258,15 @@ PUT /api/mcp/config
 Content-Type: application/json
 ```
 
+Requires an authenticated admin session. API-managed `stdio` MCP servers may
+only use allowed executable names for `command` (default: `npx`, `uvx`). Set
+`DEER_FLOW_MCP_STDIO_COMMAND_ALLOWLIST` to a comma-separated list when a
+deployment needs additional trusted launchers.
+
 **Request Body:**
 ```json
 {
-  "mcpServers": {
+  "mcp_servers": {
     "github": {
       "enabled": true,
       "type": "stdio",
@@ -276,8 +284,18 @@ Content-Type: application/json
 **Response:**
 ```json
 {
-  "success": true,
-  "message": "MCP configuration updated"
+  "mcp_servers": {
+    "github": {
+      "enabled": true,
+      "type": "stdio",
+      "command": "npx",
+      "args": ["-y", "@modelcontextprotocol/server-github"],
+      "env": {
+        "GITHUB_TOKEN": "***"
+      },
+      "description": "GitHub operations"
+    }
+  }
 }
 ```
 

backend/docs/CONFIGURATION.md

@@ -113,7 +113,7 @@ models:
     base_url: https://api.minimax.io/v1
     max_tokens: 4096
     temperature: 1.0  # MiniMax requires temperature in (0.0, 1.0]
-    supports_vision: true
+    supports_vision: false  # M2.7 is text-only; M3 supports vision
 
   - name: minimax-m2.7-highspeed
     display_name: MiniMax M2.7 Highspeed
@@ -123,7 +123,7 @@ models:
     base_url: https://api.minimax.io/v1
     max_tokens: 4096
     temperature: 1.0  # MiniMax requires temperature in (0.0, 1.0]
-    supports_vision: true
+    supports_vision: false  # M2.7 is text-only; M3 supports vision
   - name: openrouter-gemini-2.5-flash
     display_name: Gemini 2.5 Flash (OpenRouter)
     use: langchain_openai:ChatOpenAI

backend/docs/IM_CHANNEL_CONNECTIONS.md

@@ -0,0 +1,84 @@
+# IM Channel Connections
+
+DeerFlow supports user-owned IM channel bindings for Telegram, Slack, and Discord. The feature reuses the existing `channels.*` runtime configuration, so it works in local and private deployments with the same outbound transports already supported by DeerFlow.
+
+No public IP, OAuth callback URL, or provider webhook is required in this implementation.
+
+## Configuration
+
+Configure the actual IM bots under the existing `channels` block:
+
+```yaml
+channels:
+  telegram:
+    enabled: true
+    bot_token: $TELEGRAM_BOT_TOKEN
+
+  slack:
+    enabled: true
+    bot_token: $SLACK_BOT_TOKEN
+    app_token: $SLACK_APP_TOKEN
+
+  discord:
+    enabled: true
+    bot_token: $DISCORD_BOT_TOKEN
+```
+
+Then enable user bindings in `channel_connections`:
+
+```yaml
+channel_connections:
+  enabled: true
+
+  telegram:
+    enabled: true
+    bot_username: $TELEGRAM_BOT_USERNAME
+
+  slack:
+    enabled: true
+
+  discord:
+    enabled: true
+```
+
+`channel_connections` does not duplicate provider secrets. It only controls the browser-facing connect UI and stores per-user binding records. Telegram needs `bot_username` only so the frontend can open a deep link.
+
+## Connect Flow
+
+Telegram:
+
+- The frontend creates a short one-time code.
+- The Connect button opens `https://t.me/<bot_username>?start=<code>`.
+- The existing Telegram long-polling worker receives `/start <code>` and binds that Telegram chat/user to the current DeerFlow user.
+
+Slack:
+
+- The frontend creates a short one-time code.
+- The UI shows `Send /connect <code> to the DeerFlow Slack bot.`
+- The existing Slack Socket Mode worker receives the message and binds the Slack user/team to the current DeerFlow user.
+
+Discord:
+
+- The frontend creates a short one-time code.
+- The UI shows `Send /connect <code> to the DeerFlow Discord bot.`
+- The existing Discord Gateway worker receives the message and binds the Discord user/guild to the current DeerFlow user.
+
+Codes expire after 10 minutes and are single-use.
+
+## Runtime Model
+
+Connection records live in SQL tables under `deerflow.persistence.channel_connections`:
+
+- `channel_connections`: owner user, provider identity, workspace/guild/team, status, metadata.
+- `channel_oauth_states`: one-time connect codes and Telegram deep-link state.
+- `channel_conversations`: connection-scoped IM conversation to DeerFlow thread mapping.
+- `channel_credentials`: reserved for future provider-token flows, not used by the local/private binding flow.
+
+Incoming messages that resolve to a connection carry `connection_id`, `owner_user_id`, and `workspace_id`. `ChannelManager` uses `owner_user_id` as the DeerFlow run user id and preserves the raw platform user id as `channel_user_id`.
+
+## Security Notes
+
+- Browser APIs remain authenticated and CSRF-protected.
+- Connect codes are random, short-lived, and single-use.
+- Provider bot tokens remain in `channels.*` and are never returned to the browser.
+- This implementation does not add public provider callback or webhook routes.

backend/docs/REPLAY_E2E.md

@@ -0,0 +1,120 @@
+# Record/Replay E2E — front-back contract verification
+
+Deterministic, **key-free** end-to-end checks that a backend change can't
+silently break the frontend (and vice-versa). Two complementary layers, fed by a
+single recording.
+
+## Why
+
+The mock-based frontend e2e hand-writes the backend's JSON/SSE, so a backend
+schema or SSE change passes green ("fake green"). These layers replay a recorded
+**real** run against the **real** backend (and, for Layer 2, the real frontend),
+so contract drift turns the build red instead.
+
+## The two layers
+
+- **Layer 1 — backend golden** (`tests/test_replay_golden.py`): replays a fixture
+  through the real FastAPI gateway with `ReplayChatModel` and asserts the streamed
+  SSE event sequence equals a committed golden. Fast, no browser. Guards protocol
+  *shape*.
+- **Layer 2 — full-stack render** (`frontend/tests/e2e-real-backend/`): real
+  Next.js + real gateway (replay model) + Chromium; asserts the replayed
+  auto-title and a follow-up suggestion render in the browser. Guards semantic
+  *render*. (Complementary to Layer 1 — neither subsumes the other.)
+
+Layer 2 also hosts **cross-stack contract scenarios** — the dangerous class
+where a backend change silently breaks a frontend assumption and *both sides'
+unit tests stay green*. See below.
+
+## Cross-stack scenario: multi-run render order (`multi-run-order.spec.ts`)
+
+Regression guard for issue **#3352** (after context compression, refreshing a
+thread rendered history out of order). Root cause was a front-back desync:
+backend `RunManager.list_by_thread` returns runs **newest-first** (PR #2932),
+while the frontend (`core/threads/hooks.ts`) iterated runs and **prepended** each
+loaded page — inverting chronological order once the checkpoint no longer held
+the older messages. The backend ordering test was green throughout, and the
+frontend regression unit test hardcodes "backend returns newest-first" in a mock,
+so only a *real frontend against a real backend* catches the desync.
+
+This scenario does **not** record a conversation. It uses a **test-only seeder**
+(`tests/seed_runs_router.py`, mounted on the replay gateway only when
+`DEERFLOW_ENABLE_TEST_SEED=1`) to stand up a thread with ≥2 runs and per-run
+message events — and deliberately **no checkpoint**, which is the #3352
+precondition: it forces the frontend's per-run reload path to be the sole source
+of truth so the ordering bug becomes observable. The seeder writes through the
+gateway's own run/event stores using the request's auth context, so the real
+`list_by_thread` → `/runs/{id}/messages` → prepend path runs live. Reverting the
+#3354 frontend fix turns this spec red.
+
+## How replay works
+
+`tests/replay_provider.py::ReplayChatModel` returns recorded assistant turns keyed
+by a **normalized hash of the model caller + conversation**. The conversation is
+human / ai / tool messages — role, text, tool-call name+args; with
+`<system-reminder>`, dates, UUIDs, tmp paths stripped. The caller is the stable
+source of the model call (`lead_agent`, `middleware:title`, `suggest_agent`,
+`subagent:*`, etc.). A miss raises loudly rather than passing silently.
+
+**The system prompt is excluded from the match key.** The lead-agent system
+prompt is a living, frequently-edited implementation detail — its wording changes
+across PRs (e.g. #3195 added a "File Editing Workflow" section). Hashing it would
+make every fixture go stale and red-fail unrelated PRs the moment anyone edits the
+prompt. The conversation flow (user input → tool calls → results → answer) is the
+stable contract that identifies a recorded turn. The caller still stays in the
+key so two different model users with identical conversation text do not compete
+for the same replay bucket. (This mirrors how open-design's mock picker keys on
+the user prompt, not the system internals.) Combined with pinning skills +
+extensions empty and disabling memory/summarization
+(`tests/_replay_fixture.py::build_config_yaml`), a fixture replays the same across
+machines, days, prompt edits, and CI. Replaying needs **no API key**.
+
+A swallowed hash-miss keeps the SSE *event shapes* identical (the gateway wraps it
+into a normal assistant error message), so the Layer-1 golden can't catch a miss
+by shape alone — it inspects `replay_provider.replay_misses()` and fails loud
+instead. Layer-2 already fails on a miss (the recorded turns never render).
+
+## Record a new scenario (needs a real key — dev machine only)
+
+Recording drives the **real frontend** so captured inputs match exactly what the
+browser sends; fixtures contain no API key.
+
+```bash
+# 1. drive the real frontend against a real-model gateway, capturing model calls
+OPENAI_API_KEY=... OPENAI_API_BASE=<openai-compatible-endpoint>/v1 \
+  DEERFLOW_RECORD_OUT=/tmp/rec/turns.jsonl RECORD_MODEL=<model> \
+  bash -c 'cd frontend && pnpm exec playwright test -c playwright.record.config.ts'
+
+# 2. stitch the capture into a fixture
+cd backend && uv run python scripts/build_fixture_from_jsonl.py \
+  --jsonl /tmp/rec/turns.jsonl --meta /tmp/rec/turns.jsonl.meta.json \
+  --out tests/fixtures/replay/<scenario>.<mode>.json --model <model>
+
+# 3. regenerate the committed golden
+DEERFLOW_WRITE_GOLDEN=1 PYTHONPATH=. uv run pytest tests/test_replay_golden.py
+```
+
+## Run (no key)
+
+```bash
+cd backend  && PYTHONPATH=. uv run pytest tests/test_replay_golden.py          # Layer 1
+cd frontend && pnpm exec playwright test -c playwright.real-backend.config.ts  # Layer 2
+```
+
+## CI
+
+`.github/workflows/replay-e2e.yml` runs both layers on changes to **either** side
+of the contract (`frontend/**`, `backend/app/gateway/**`,
+`backend/packages/harness/**`, fixtures). DOM assertions are the gate; the rendered
+screenshot + Playwright HTML report are uploaded as a CI artifact.
+
+## Known limitations
+
+- Visual regression baselines are OS-specific, so they are a **local dev gate
+  only** (gitignored); CI uploads the render as an artifact for human review
+  instead of hard-asserting a cross-OS baseline.
+- Fixtures are coupled to the recording-time prompt; if new
+  environment-dependent content enters the system prompt, extend the
+  normalization in `replay_provider.py` (or pin it in `build_config_yaml`).
+- Re-record a scenario if the agent graph changes how many model calls it makes
+  — the replay raises loudly on a hash miss pointing at the divergence.

backend/docs/plan_mode_usage.md

@@ -127,8 +127,8 @@ complex_agent = create_agent_for_task("high")
 ## How It Works
 
 1. When `make_lead_agent(config)` is called, it extracts `is_plan_mode` from `config.configurable`
-2. The config is passed to `_build_middlewares(config)`
-3. `_build_middlewares()` reads `is_plan_mode` and calls `_create_todo_list_middleware(is_plan_mode)`
+2. The config is passed to `build_middlewares(config)`
+3. `build_middlewares()` reads `is_plan_mode` and calls `_create_todo_list_middleware(is_plan_mode)`
 4. If `is_plan_mode=True`, a `TodoListMiddleware` instance is created and added to the middleware chain
 5. The middleware automatically adds a `write_todos` tool to the agent's toolset
 6. The agent can use this tool to manage tasks during execution
@@ -141,7 +141,7 @@ make_lead_agent(config)
   │
   ├─> Extracts: is_plan_mode = config.configurable.get("is_plan_mode", False)
   │
-  └─> _build_middlewares(config)
+  └─> build_middlewares(config)
         │
         ├─> ThreadDataMiddleware
         ├─> SandboxMiddleware
@@ -156,7 +156,7 @@ make_lead_agent(config)
 ### Agent Module
 - **Location**: `packages/harness/deerflow/agents/lead_agent/agent.py`
 - **Function**: `_create_todo_list_middleware(is_plan_mode: bool)` - Creates TodoListMiddleware if plan mode is enabled
-- **Function**: `_build_middlewares(config: RunnableConfig)` - Builds middleware chain based on runtime config
+- **Function**: `build_middlewares(config: RunnableConfig)` - Builds middleware chain based on runtime config
 - **Function**: `make_lead_agent(config: RunnableConfig)` - Creates agent with appropriate middlewares
 
 ### Runtime Configuration

backend/packages/harness/deerflow/agents/lead_agent/agent.py

@@ -21,7 +21,6 @@ middleware, and the async path inside ``TitleMiddleware``. Any new in-graph
 from __future__ import annotations
 
 import logging
-from typing import TYPE_CHECKING
 
 from langchain.agents import create_agent
 from langchain.agents.middleware import AgentMiddleware
@@ -48,13 +47,10 @@ from deerflow.skills.tool_policy import filter_tools_by_skill_allowed_tools
 from deerflow.skills.types import Skill
 from deerflow.tracing import build_tracing_callbacks
 
-if TYPE_CHECKING:
-    from langchain.tools import BaseTool
-
-    from deerflow.tools.builtins.tool_search import DeferredToolSetup
-
 logger = logging.getLogger(__name__)
 
+_BOOTSTRAP_SKILL_NAMES = {"bootstrap"}
+
 
 def _get_runtime_config(config: RunnableConfig) -> dict:
     """Merge legacy configurable options with LangGraph runtime context."""
@@ -271,21 +267,31 @@ Being proactive with task management demonstrates thoroughness and ensures all r
 # ViewImageMiddleware should be before ClarificationMiddleware to inject image details before LLM
 # ToolErrorHandlingMiddleware should be before ClarificationMiddleware to convert tool exceptions to ToolMessages
 # ClarificationMiddleware should be last to intercept clarification requests after model calls
-def _build_middlewares(
+def build_middlewares(
     config: RunnableConfig,
     model_name: str | None,
     agent_name: str | None = None,
     custom_middlewares: list[AgentMiddleware] | None = None,
     *,
+    available_skills: set[str] | None = None,
     app_config: AppConfig | None = None,
     deferred_setup=None,
 ):
-    """Build middleware chain based on runtime configuration.
+    """Build the lead-agent middleware chain based on runtime configuration.
+
+    Public entry point for the lead agent's full middleware composition. Used by
+    ``make_lead_agent`` and by the embedded ``DeerFlowClient`` (a lead-agent variant
+    that needs the identical chain). Keep this name stable: it is imported across a
+    module boundary, so renames/signature changes ripple into ``client.py``.
 
     Args:
         config: Runtime configuration containing configurable options like is_plan_mode.
+        model_name: Resolved runtime model name; gates vision-only middleware.
         agent_name: If provided, MemoryMiddleware will use per-agent memory storage.
         custom_middlewares: Optional list of custom middlewares to inject into the chain.
+        app_config: Explicit AppConfig; falls back to ``get_app_config()`` when omitted.
+        deferred_setup: Optional deferred-MCP-tool setup that attaches
+            ``DeferredToolFilterMiddleware`` when ``tool_search`` is enabled.
 
     Returns:
         List of middleware instances.
@@ -299,6 +305,13 @@ def _build_middlewares(
 
     middlewares.append(DynamicContextMiddleware(agent_name=agent_name, app_config=resolved_app_config))
 
+    # Deterministically load a full SKILL.md when the user starts the turn with
+    # /skill-name. This keeps the base system prompt metadata-only while giving
+    # explicit user activation priority over model-side relevance guessing.
+    from deerflow.agents.middlewares.skill_activation_middleware import SkillActivationMiddleware
+
+    middlewares.append(SkillActivationMiddleware(available_skills=available_skills, app_config=resolved_app_config))
+
     # Add summarization middleware if enabled
     summarization_middleware = _create_summarization_middleware(app_config=resolved_app_config)
     if summarization_middleware is not None:
@@ -364,29 +377,9 @@ def _build_middlewares(
     return middlewares
 
 
-def _assemble_deferred(filtered_tools: list[BaseTool], *, enabled: bool) -> tuple[list[BaseTool], DeferredToolSetup]:
-    """Build the final tool list + deferred setup from a policy-filtered list.
-
-    Call AFTER tool-policy filtering so the deferred catalog never exposes a
-    tool the agent is not allowed to use. Fail-closed: if tool_search is enabled
-    and MCP tools survived filtering but no deferred set was recovered, raise
-    rather than silently binding their full schemas to the model.
-    """
-    from deerflow.tools.builtins.tool_search import build_deferred_tool_setup
-    from deerflow.tools.mcp_metadata import is_mcp_tool
-
-    deferred_setup = build_deferred_tool_setup(filtered_tools, enabled=enabled)
-    if enabled and not deferred_setup.deferred_names and any(is_mcp_tool(t) for t in filtered_tools):
-        raise RuntimeError("tool_search enabled and MCP tools survived policy filtering, but no deferred set was recovered — refusing to bind MCP schemas (fail-closed).")
-    final_tools = list(filtered_tools)
-    if deferred_setup.tool_search_tool:
-        final_tools.append(deferred_setup.tool_search_tool)
-    return final_tools, deferred_setup
-
-
 def _available_skill_names(agent_config, is_bootstrap: bool) -> set[str] | None:
     if is_bootstrap:
-        return {"bootstrap"}
+        return set(_BOOTSTRAP_SKILL_NAMES)
     if agent_config and agent_config.skills is not None:
         return set(agent_config.skills)
     return None
@@ -417,6 +410,7 @@ def _make_lead_agent(config: RunnableConfig, *, app_config: AppConfig):
     # Lazy import to avoid circular dependency
     from deerflow.tools import get_available_tools
     from deerflow.tools.builtins import setup_agent, update_agent
+    from deerflow.tools.builtins.tool_search import assemble_deferred_tools
 
     cfg = _get_runtime_config(config)
     resolved_app_config = app_config
@@ -491,17 +485,25 @@ def _make_lead_agent(config: RunnableConfig, *, app_config: AppConfig):
 
     if is_bootstrap:
         # Special bootstrap agent with minimal prompt for initial custom agent creation flow
+        # Keep the bootstrap skill set intentionally narrow so agent creation
+        # remains deterministic before the custom agent's own config exists.
         raw_tools = get_available_tools(model_name=model_name, subagent_enabled=subagent_enabled, app_config=resolved_app_config) + [setup_agent]
         filtered = filter_tools_by_skill_allowed_tools(raw_tools, skills_for_tool_policy)
-        final_tools, setup = _assemble_deferred(filtered, enabled=resolved_app_config.tool_search.enabled)
+        final_tools, setup = assemble_deferred_tools(filtered, enabled=resolved_app_config.tool_search.enabled)
         return create_agent(
             model=create_chat_model(name=model_name, thinking_enabled=thinking_enabled, app_config=resolved_app_config, attach_tracing=False),
             tools=final_tools,
-            middleware=_build_middlewares(config, model_name=model_name, app_config=resolved_app_config, deferred_setup=setup),
+            middleware=build_middlewares(
+                config,
+                model_name=model_name,
+                available_skills=set(_BOOTSTRAP_SKILL_NAMES),
+                app_config=resolved_app_config,
+                deferred_setup=setup,
+            ),
             system_prompt=apply_prompt_template(
                 subagent_enabled=subagent_enabled,
                 max_concurrent_subagents=max_concurrent_subagents,
-                available_skills=set(["bootstrap"]),
+                available_skills=set(_BOOTSTRAP_SKILL_NAMES),
                 app_config=resolved_app_config,
                 deferred_names=setup.deferred_names,
             ),
@@ -514,16 +516,23 @@ def _make_lead_agent(config: RunnableConfig, *, app_config: AppConfig):
     # Default lead agent (unchanged behavior)
     raw_tools = get_available_tools(model_name=model_name, groups=agent_config.tool_groups if agent_config else None, subagent_enabled=subagent_enabled, app_config=resolved_app_config)
     filtered = filter_tools_by_skill_allowed_tools(raw_tools + extra_tools, skills_for_tool_policy)
-    final_tools, setup = _assemble_deferred(filtered, enabled=resolved_app_config.tool_search.enabled)
+    final_tools, setup = assemble_deferred_tools(filtered, enabled=resolved_app_config.tool_search.enabled)
     return create_agent(
         model=create_chat_model(name=model_name, thinking_enabled=thinking_enabled, reasoning_effort=reasoning_effort, app_config=resolved_app_config, attach_tracing=False),
         tools=final_tools,
-        middleware=_build_middlewares(config, model_name=model_name, agent_name=agent_name, app_config=resolved_app_config, deferred_setup=setup),
+        middleware=build_middlewares(
+            config,
+            model_name=model_name,
+            agent_name=agent_name,
+            available_skills=available_skills,
+            app_config=resolved_app_config,
+            deferred_setup=setup,
+        ),
         system_prompt=apply_prompt_template(
             subagent_enabled=subagent_enabled,
             max_concurrent_subagents=max_concurrent_subagents,
             agent_name=agent_name,
-            available_skills=set(agent_config.skills) if agent_config and agent_config.skills is not None else None,
+            available_skills=available_skills,
             app_config=resolved_app_config,
             deferred_names=setup.deferred_names,
         ),

backend/packages/harness/deerflow/agents/lead_agent/prompt.py

@@ -10,6 +10,7 @@ from deerflow.config.agents_config import load_agent_soul
 from deerflow.skills.storage import get_or_new_skill_storage
 from deerflow.skills.types import Skill, SkillCategory
 from deerflow.subagents import get_available_subagent_names
+from deerflow.tools.builtins.tool_search import get_deferred_tools_prompt_section
 
 if TYPE_CHECKING:
     from deerflow.config.app_config import AppConfig
@@ -624,6 +625,11 @@ You have access to skills that provide optimized workflows for specific tasks. E
 4. Load referenced resources only when needed during execution
 5. Follow the skill's instructions precisely
 
+**Explicit Slash Skill Activation:**
+- If the user starts a request with `/<skill-name>`, that skill was explicitly requested for the current turn.
+- Follow the activated skill before choosing a general workflow.
+- The runtime injects the activated skill content for explicit slash activations; do not call `read_file` for that SKILL.md again unless the injected skill references supporting resources you need.
+
 **Skills are located at:** {container_base_path}
 {skill_evolution_section}
 {skills_list}
@@ -693,19 +699,6 @@ Rules:
 """
 
 
-def get_deferred_tools_prompt_section(*, deferred_names: frozenset[str] = frozenset()) -> str:
-    """Generate <available-deferred-tools> from an explicit deferred-name set.
-
-    Lists only names so the agent knows what exists and can use tool_search to
-    load them. Returns empty string when there are no deferred tools. The set is
-    computed at agent build time (after tool-policy filtering) and passed in.
-    """
-    if not deferred_names:
-        return ""
-    names = "\n".join(sorted(deferred_names))
-    return f"<available-deferred-tools>\n{names}\n</available-deferred-tools>"
-
-
 def _build_acp_section(*, app_config: AppConfig | None = None) -> str:
     """Build the ACP agent prompt section, only if ACP agents are configured."""
     if app_config is None:

backend/packages/harness/deerflow/agents/memory/prompt.py

@@ -1,9 +1,14 @@
 """Prompt templates for memory update and injection."""
 
+from __future__ import annotations
+
+import logging
 import math
 import re
 from typing import Any
 
+logger = logging.getLogger(__name__)
+
 try:
     import tiktoken
 
@@ -160,6 +165,39 @@ Rules:
 Return ONLY valid JSON."""
 
 
+# Module-level tiktoken encoding cache.  Populated lazily on first use;
+# subsequent calls are a dict lookup (no network I/O).  Pre-warming at
+# startup via :func:`warm_tiktoken_cache` avoids blocking a request on the
+# (potentially slow) first ``get_encoding`` call.
+_tiktoken_encoding_cache: dict[str, tiktoken.Encoding] = {}
+
+
+def _get_tiktoken_encoding(encoding_name: str = "cl100k_base") -> tiktoken.Encoding | None:
+    """Return a cached tiktoken encoding, or ``None`` on failure / unavailability.
+
+    On the very first call for a given *encoding_name*, tiktoken may need to
+    download the BPE data from ``openaipublic.blob.core.windows.net``.  In
+    network-restricted environments (e.g. deployments behind the GFW) this
+    download can block for tens of minutes before the OS TCP timeout kicks in.
+    The caller must therefore be prepared for this to block and should run it
+    off the event loop (e.g. via ``asyncio.to_thread``).
+    """
+    if not TIKTOKEN_AVAILABLE:
+        return None
+
+    cached = _tiktoken_encoding_cache.get(encoding_name)
+    if cached is not None:
+        return cached
+
+    try:
+        encoding = tiktoken.get_encoding(encoding_name)
+        _tiktoken_encoding_cache[encoding_name] = encoding
+        return encoding
+    except Exception:
+        logger.warning("Failed to load tiktoken encoding %r; falling back to char-based estimation", encoding_name, exc_info=True)
+        return None
+
+
 def _count_tokens(text: str, encoding_name: str = "cl100k_base") -> int:
     """Count tokens in text using tiktoken.
 
@@ -170,18 +208,30 @@ def _count_tokens(text: str, encoding_name: str = "cl100k_base") -> int:
     Returns:
         The number of tokens in the text.
     """
-    if not TIKTOKEN_AVAILABLE:
+    encoding = _get_tiktoken_encoding(encoding_name)
+    if encoding is None:
         # Fallback to character-based estimation if tiktoken is not available
+        # or the encoding failed to load.
         return len(text) // 4
 
     try:
-        encoding = tiktoken.get_encoding(encoding_name)
         return len(encoding.encode(text))
     except Exception:
         # Fallback to character-based estimation on error
         return len(text) // 4
 
 
+def warm_tiktoken_cache() -> bool:
+    """Pre-warm the tiktoken encoding cache.
+
+    Call at startup (off the event loop) so the first request never blocks
+    on the BPE download.  Returns ``True`` if the encoding was loaded
+    successfully (or was already cached), ``False`` if tiktoken is
+    unavailable or the download failed.
+    """
+    return _get_tiktoken_encoding("cl100k_base") is not None
+
+
 def _coerce_confidence(value: Any, default: float = 0.0) -> float:
     """Coerce a confidence-like value to a bounded float in [0, 1].
 

backend/packages/harness/deerflow/agents/middlewares/dynamic_context_middleware.py

@@ -28,6 +28,7 @@ Date-update format:
 
 from __future__ import annotations
 
+import asyncio
 import logging
 import re
 import uuid
@@ -43,6 +44,12 @@ if TYPE_CHECKING:
 
 logger = logging.getLogger(__name__)
 
+# Upper bound (seconds) for a single _inject() offload.  If the warm-up at
+# gateway startup failed silently, the first request may still hit a cold
+# tiktoken BPE download that blocks until the OS TCP timeout (~26 min).
+# This cap ensures the request degrades gracefully instead of hanging.
+_INJECT_TIMEOUT_SECONDS = 5.0
+
 _DATE_RE = re.compile(r"<current_date>([^<]+)</current_date>")
 _DYNAMIC_CONTEXT_REMINDER_KEY = "dynamic_context_reminder"
 _SUMMARY_MESSAGE_NAME = "summary"
@@ -201,4 +208,25 @@ class DynamicContextMiddleware(AgentMiddleware):
 
     @override
     async def abefore_agent(self, state, runtime: Runtime) -> dict | None:
-        return self._inject(state)
+        # _inject() performs synchronous file I/O (memory JSON loading) and
+        # potentially blocking network calls (tiktoken encoding download on
+        # first use).  Offload to a thread so the event loop is never blocked
+        # — a blocking call here starves all concurrent HTTP handlers (auth,
+        # SSE heartbeats, etc.).  See issue #3402.
+        #
+        # Bounded timeout: if startup warm-up failed silently (e.g. network
+        # blip during deploy), the first request's cold tiktoken download can
+        # block for tens of minutes (OS TCP timeout).  Time-box injection so
+        # the request degrades gracefully (no memory context) rather than
+        # hanging.
+        try:
+            return await asyncio.wait_for(
+                asyncio.to_thread(self._inject, state),
+                timeout=_INJECT_TIMEOUT_SECONDS,
+            )
+        except TimeoutError:
+            logger.warning(
+                "DynamicContextMiddleware: injection timed out (%.1fs); skipping memory/date injection for this turn",
+                _INJECT_TIMEOUT_SECONDS,
+            )
+            return None

backend/packages/harness/deerflow/agents/middlewares/skill_activation_middleware.py

@@ -0,0 +1,289 @@
+"""Middleware for explicit slash skill activation."""
+
+from __future__ import annotations
+
+import asyncio
+import hashlib
+import html
+import logging
+import uuid
+from collections.abc import Awaitable, Callable
+from dataclasses import dataclass
+from pathlib import Path
+from typing import TYPE_CHECKING, override
+
+from langchain.agents.middleware import AgentMiddleware
+from langchain.agents.middleware.types import ModelRequest, ModelResponse
+from langchain_core.messages import AIMessage, HumanMessage
+
+from deerflow.skills.slash import parse_slash_skill_reference, resolve_slash_skill
+from deerflow.skills.storage import get_or_new_skill_storage
+from deerflow.skills.storage.skill_storage import SkillStorage
+from deerflow.skills.types import SKILL_MD_FILE
+from deerflow.utils.messages import get_original_user_content_text
+
+if TYPE_CHECKING:
+    from deerflow.config.app_config import AppConfig
+
+logger = logging.getLogger(__name__)
+
+_SLASH_SKILL_ACTIVATION_KEY = "slash_skill_activation"
+_SLASH_SKILL_ACTIVATION_TARGET_ID_KEY = "slash_skill_activation_target_id"
+_SUMMARY_MESSAGE_NAME = "summary"
+
+
+@dataclass(frozen=True, slots=True)
+class _Activation:
+    skill_name: str
+    category: str
+    container_file_path: str
+    skill_content: str
+    content_hash: str
+    remaining_text: str
+
+
+@dataclass(frozen=True, slots=True)
+class _ActivationResolution:
+    activation: _Activation | None = None
+    failure_message: str | None = None
+
+
+def is_slash_skill_activation_reminder(message: object) -> bool:
+    """Return whether a message is hidden slash-skill activation context."""
+    return isinstance(message, HumanMessage) and bool(message.additional_kwargs.get(_SLASH_SKILL_ACTIVATION_KEY))
+
+
+def _is_user_activation_target(message: object) -> bool:
+    if not isinstance(message, HumanMessage):
+        return False
+    if message.name == _SUMMARY_MESSAGE_NAME:
+        return False
+    if message.additional_kwargs.get("hide_from_ui"):
+        return False
+    return True
+
+
+class SkillActivationMiddleware(AgentMiddleware):
+    """Inject full SKILL.md content when the user explicitly types /skill-name."""
+
+    def __init__(
+        self,
+        *,
+        available_skills: set[str] | None = None,
+        app_config: AppConfig | None = None,
+    ) -> None:
+        super().__init__()
+        self._available_skills = set(available_skills) if available_skills is not None else None
+        self._app_config = app_config
+
+    def _storage(self) -> SkillStorage:
+        if self._app_config is not None:
+            return get_or_new_skill_storage(app_config=self._app_config)
+        return get_or_new_skill_storage()
+
+    @staticmethod
+    def _read_skill_content(skill_file: Path, skills_root: Path) -> str:
+        if skill_file.name != SKILL_MD_FILE:
+            raise ValueError(f"Expected {SKILL_MD_FILE}, got {skill_file.name}")
+        resolved_root = skills_root.resolve()
+        resolved_file = skill_file.resolve()
+        try:
+            resolved_file.relative_to(resolved_root)
+        except ValueError as exc:
+            raise ValueError("Resolved skill file must stay within the configured skills root.") from exc
+        if not resolved_file.is_file():
+            raise FileNotFoundError(resolved_file)
+        return resolved_file.read_text(encoding="utf-8")
+
+    def _resolve_activation(self, text: str) -> _ActivationResolution | None:
+        reference = parse_slash_skill_reference(text)
+        if reference is None:
+            return None
+
+        storage = self._storage()
+        skills = storage.load_skills(enabled_only=False)
+        skill = next((candidate for candidate in skills if candidate.name == reference.name), None)
+        if skill is None:
+            return _ActivationResolution(failure_message=f"Skill `/{reference.name}` is not installed.")
+        if not skill.enabled:
+            return _ActivationResolution(failure_message=f"Skill `/{reference.name}` is installed but disabled. Enable it before using slash activation.")
+        if self._available_skills is not None and reference.name not in self._available_skills:
+            return _ActivationResolution(failure_message=f"Skill `/{reference.name}` is not available for this agent.")
+
+        resolved = resolve_slash_skill(
+            text,
+            skills,
+            available_skills=self._available_skills,
+            container_base_path=storage.get_container_root(),
+        )
+        if resolved is None:
+            return _ActivationResolution(failure_message=f"Skill `/{reference.name}` could not be resolved.")
+
+        try:
+            skill_content = self._read_skill_content(resolved.skill.skill_file, storage.get_skills_root_path())
+        except (OSError, ValueError):
+            logger.exception("Failed to read slash-activated skill %s", resolved.skill.name)
+            return _ActivationResolution(failure_message=f"Skill `/{reference.name}` could not be loaded safely. Please check the skill installation.")
+
+        content_hash = hashlib.sha256(skill_content.encode("utf-8")).hexdigest()
+        return _ActivationResolution(
+            activation=_Activation(
+                skill_name=resolved.skill.name,
+                category=str(resolved.skill.category),
+                container_file_path=resolved.container_file_path,
+                skill_content=skill_content,
+                content_hash=content_hash,
+                remaining_text=resolved.remaining_text,
+            )
+        )
+
+    @staticmethod
+    def _build_activation_reminder(activation: _Activation) -> str:
+        user_request = activation.remaining_text or ("No additional task text was provided after the slash skill command. Ask the user what they want to do with this skill if the next step is unclear.")
+        escaped_user_request = html.escape(user_request, quote=False)
+        escaped_skill_content = html.escape(activation.skill_content, quote=False)
+        escaped_skill_name = html.escape(activation.skill_name, quote=True)
+        escaped_category = html.escape(activation.category, quote=True)
+        escaped_path = html.escape(activation.container_file_path, quote=True)
+        escaped_content_hash = html.escape(activation.content_hash, quote=True)
+        return f"""<slash_skill_activation>
+The user explicitly activated the `{activation.skill_name}` skill for this turn.
+Treat the task text as:
+<user_request>
+{escaped_user_request}
+</user_request>
+
+Follow this skill before choosing a general workflow. Load supporting resources from the same skill directory only when needed.
+
+<skill name="{escaped_skill_name}" category="{escaped_category}" path="{escaped_path}" sha256="{escaped_content_hash}">
+<skill_content encoding="xml-escaped">
+{escaped_skill_content}
+</skill_content>
+</skill>
+</slash_skill_activation>"""
+
+    @staticmethod
+    def _has_existing_activation_for_target(messages: list, target_index: int, target: HumanMessage) -> bool:
+        if target_index <= 0:
+            return False
+
+        if target.id:
+            for previous in messages[:target_index]:
+                if not is_slash_skill_activation_reminder(previous):
+                    continue
+                target_id = previous.additional_kwargs.get(_SLASH_SKILL_ACTIVATION_TARGET_ID_KEY)
+                if target_id == target.id or previous.id == f"{target.id}__slash_activation":
+                    return True
+
+        previous = messages[target_index - 1]
+        return is_slash_skill_activation_reminder(previous)
+
+    def _find_activation_target(self, messages: list) -> tuple[int, HumanMessage, _ActivationResolution] | None:
+        if not messages:
+            return None
+
+        target_index = next((idx for idx in range(len(messages) - 1, -1, -1) if _is_user_activation_target(messages[idx])), None)
+        if target_index is None:
+            return None
+
+        target = messages[target_index]
+        if target is None:
+            return None
+        if self._has_existing_activation_for_target(messages, target_index, target):
+            return None
+
+        content = get_original_user_content_text(target.content, target.additional_kwargs)
+        resolution = self._resolve_activation(content)
+        if resolution is None:
+            return None
+        return target_index, target, resolution
+
+    @staticmethod
+    def _record_activation(request: ModelRequest, activation: _Activation, *, hook: str) -> None:
+        runtime = getattr(request, "runtime", None)
+        context = getattr(runtime, "context", None)
+        journal = context.get("__run_journal") if isinstance(context, dict) else None
+        if journal is None:
+            return
+        try:
+            journal.record_middleware(
+                "skill_activation",
+                name="SkillActivationMiddleware",
+                hook=hook,
+                action="activate",
+                changes={
+                    "skill_name": activation.skill_name,
+                    "category": activation.category,
+                    "path": activation.container_file_path,
+                    "content_hash": activation.content_hash,
+                },
+            )
+        except Exception:
+            logger.debug("Failed to record slash skill activation audit event", exc_info=True)
+
+    def _prepare_model_request(self, request: ModelRequest, *, hook: str) -> ModelRequest | AIMessage | None:
+        target_and_resolution = self._find_activation_target(list(request.messages))
+        if target_and_resolution is None:
+            return None
+
+        target_index, target, resolution = target_and_resolution
+        if resolution.failure_message:
+            return AIMessage(content=resolution.failure_message)
+
+        activation = resolution.activation
+        if activation is None:
+            return None
+
+        logger.info(
+            "SkillActivationMiddleware: activating slash skill %s category=%s path=%s hash=%s",
+            activation.skill_name,
+            activation.category,
+            activation.container_file_path,
+            activation.content_hash,
+        )
+        self._record_activation(request, activation, hook=hook)
+        activation_msg = self._make_activation_message(target, self._build_activation_reminder(activation))
+        messages = list(request.messages)
+        messages.insert(target_index, activation_msg)
+        return request.override(messages=messages)
+
+    @staticmethod
+    def _make_activation_message(target: HumanMessage, activation_content: str) -> HumanMessage:
+        stable_id = target.id or str(uuid.uuid4())
+        additional_kwargs = {
+            "hide_from_ui": True,
+            _SLASH_SKILL_ACTIVATION_KEY: True,
+        }
+        if target.id:
+            additional_kwargs[_SLASH_SKILL_ACTIVATION_TARGET_ID_KEY] = target.id
+        return HumanMessage(
+            content=activation_content,
+            id=f"{stable_id}__slash_activation",
+            additional_kwargs=additional_kwargs,
+        )
+
+    @override
+    def wrap_model_call(
+        self,
+        request: ModelRequest,
+        handler: Callable[[ModelRequest], ModelResponse],
+    ) -> ModelResponse | AIMessage:
+        prepared = self._prepare_model_request(request, hook="wrap_model_call")
+        if prepared is None:
+            return handler(request)
+        if isinstance(prepared, AIMessage):
+            return prepared
+        return handler(prepared)
+
+    @override
+    async def awrap_model_call(
+        self,
+        request: ModelRequest,
+        handler: Callable[[ModelRequest], Awaitable[ModelResponse]],
+    ) -> ModelResponse | AIMessage:
+        prepared = await asyncio.to_thread(self._prepare_model_request, request, hook="awrap_model_call")
+        if prepared is None:
+            return await handler(request)
+        if isinstance(prepared, AIMessage):
+            return prepared
+        return await handler(prepared)

backend/packages/harness/deerflow/agents/middlewares/tool_error_handling_middleware.py

@@ -2,7 +2,7 @@
 
 import logging
 from collections.abc import Awaitable, Callable
-from typing import override
+from typing import TYPE_CHECKING, override
 
 from langchain.agents import AgentState
 from langchain.agents.middleware import AgentMiddleware
@@ -12,10 +12,48 @@ from langgraph.prebuilt.tool_node import ToolCallRequest
 from langgraph.types import Command
 
 from deerflow.config.app_config import AppConfig
+from deerflow.subagents.status_contract import (
+    extract_subagent_status,
+    make_subagent_additional_kwargs,
+)
+
+if TYPE_CHECKING:
+    from deerflow.tools.builtins.tool_search import DeferredToolSetup
 
 logger = logging.getLogger(__name__)
 
 _MISSING_TOOL_CALL_ID = "missing_tool_call_id"
+_TASK_TOOL_NAME = "task"
+
+
+def _stamp_task_subagent_status(message: ToolMessage, *, tool_name: str, error: str | None = None) -> ToolMessage:
+    """Centralised stamping of ``additional_kwargs.subagent_status``.
+
+    Bytedance/deer-flow issue #3146: the frontend now reads the subagent
+    status from a structured field instead of parsing the leading text of
+    the task tool's return string. That contract is enforced here, in the
+    one place every task tool result flows through, rather than at the 5
+    normal-return + 3 ``Error:`` pre-execution branches inside
+    ``task_tool.py``. Centralisation prevents the "added a new return
+    path, forgot the stamp" drift mode.
+
+    For non-``task`` tools this is a no-op so other tools' additional_kwargs
+    conventions are untouched.
+    """
+    if tool_name != _TASK_TOOL_NAME:
+        return message
+    content = message.content if isinstance(message.content, str) else ""
+    status = extract_subagent_status(content)
+    if status is None:
+        # Non-terminal streaming chunks or unrecognised shapes leave the
+        # field unset so the frontend can keep the card on its in-progress
+        # placeholder until a real terminal frame arrives.
+        return message
+    stamp = make_subagent_additional_kwargs(status, error=error)
+    existing = dict(message.additional_kwargs or {})
+    existing.update(stamp)
+    message.additional_kwargs = existing
+    return message
 
 
 class ToolErrorHandlingMiddleware(AgentMiddleware[AgentState]):
@@ -29,12 +67,31 @@ class ToolErrorHandlingMiddleware(AgentMiddleware[AgentState]):
             detail = detail[:497] + "..."
 
         content = f"Error: Tool '{tool_name}' failed with {exc.__class__.__name__}: {detail}. Continue with available context, or choose an alternative tool."
-        return ToolMessage(
+        message = ToolMessage(
             content=content,
             tool_call_id=tool_call_id,
             name=tool_name,
             status="error",
         )
+        # Stamp the structured subagent status on the wrapper too: the
+        # frontend would otherwise have to fall back to prefix-matching
+        # ``Error: Tool 'task' failed ...`` on the wire. The ``subagent_error``
+        # carries the same ``ExcClass: detail`` shape the wrapper string
+        # uses so debugging artifacts stay aligned.
+        structured_error = f"{exc.__class__.__name__}: {detail}"
+        return _stamp_task_subagent_status(message, tool_name=tool_name, error=structured_error)
+
+    @staticmethod
+    def _maybe_stamp(result: ToolMessage | Command, request: ToolCallRequest) -> ToolMessage | Command:
+        """Apply the subagent stamp to successful task tool returns.
+
+        ``Command`` results bypass the stamp — they encode LangGraph
+        control flow rather than user-facing tool output.
+        """
+        if not isinstance(result, ToolMessage):
+            return result
+        tool_name = str(request.tool_call.get("name") or "")
+        return _stamp_task_subagent_status(result, tool_name=tool_name)
 
     @override
     def wrap_tool_call(
@@ -43,13 +100,14 @@ class ToolErrorHandlingMiddleware(AgentMiddleware[AgentState]):
         handler: Callable[[ToolCallRequest], ToolMessage | Command],
     ) -> ToolMessage | Command:
         try:
-            return handler(request)
+            result = handler(request)
         except GraphBubbleUp:
             # Preserve LangGraph control-flow signals (interrupt/pause/resume).
             raise
         except Exception as exc:
             logger.exception("Tool execution failed (sync): name=%s id=%s", request.tool_call.get("name"), request.tool_call.get("id"))
             return self._build_error_message(request, exc)
+        return self._maybe_stamp(result, request)
 
     @override
     async def awrap_tool_call(
@@ -58,13 +116,14 @@ class ToolErrorHandlingMiddleware(AgentMiddleware[AgentState]):
         handler: Callable[[ToolCallRequest], Awaitable[ToolMessage | Command]],
     ) -> ToolMessage | Command:
         try:
-            return await handler(request)
+            result = await handler(request)
         except GraphBubbleUp:
             # Preserve LangGraph control-flow signals (interrupt/pause/resume).
             raise
         except Exception as exc:
             logger.exception("Tool execution failed (async): name=%s id=%s", request.tool_call.get("name"), request.tool_call.get("id"))
             return self._build_error_message(request, exc)
+        return self._maybe_stamp(result, request)
 
 
 def _build_runtime_middlewares(
@@ -143,6 +202,7 @@ def build_subagent_runtime_middlewares(
     app_config: AppConfig | None = None,
     model_name: str | None = None,
     lazy_init: bool = True,
+    deferred_setup: "DeferredToolSetup | None" = None,
 ) -> list[AgentMiddleware]:
     """Middlewares shared by subagent runtime before subagent-only middlewares."""
     if app_config is None:
@@ -166,6 +226,16 @@ def build_subagent_runtime_middlewares(
 
         middlewares.append(ViewImageMiddleware())
 
+    # Hide deferred (MCP) tool schemas from the subagent's model binding until
+    # tool_search promotes them. This is the same wiring the lead agent gets. The deferred
+    # set + catalog hash come from the build-time setup (assembled after
+    # tool-policy filtering); promotion is read from graph state. Empty/None
+    # setup (deferral disabled or no MCP tool survived) is a pure no-op.
+    if deferred_setup is not None and deferred_setup.deferred_names:
+        from deerflow.agents.middlewares.deferred_tool_filter_middleware import DeferredToolFilterMiddleware
+
+        middlewares.append(DeferredToolFilterMiddleware(deferred_setup.deferred_names, deferred_setup.catalog_hash))
+
     # Same provider safety-termination guard the lead agent uses — subagents
     # are equally exposed to truncated tool_calls returned with
     # finish_reason=content_filter (and friends), and the bad call would then

backend/packages/harness/deerflow/agents/middlewares/tool_output_budget_middleware.py

@@ -11,10 +11,11 @@ from __future__ import annotations
 import asyncio
 import logging
 import os
+import shlex
 import uuid
 from collections.abc import Awaitable, Callable
 from dataclasses import replace as dc_replace
-from typing import Any, override
+from typing import TYPE_CHECKING, Any, override
 
 from langchain.agents import AgentState
 from langchain.agents.middleware import AgentMiddleware
@@ -24,9 +25,19 @@ from langgraph.prebuilt.tool_node import ToolCallRequest
 from langgraph.types import Command
 
 from deerflow.config.tool_output_config import ToolOutputConfig
+from deerflow.sandbox.sandbox_provider import get_sandbox_provider
+
+if TYPE_CHECKING:
+    from deerflow.sandbox.sandbox import Sandbox
 
 logger = logging.getLogger(__name__)
 
+# Virtual outputs root inside the sandbox. Host-mounted sandboxes map this to
+# the thread outputs dir on the host; for non-mounted (remote) sandboxes the
+# same path is written directly into the sandbox filesystem so the model's
+# ``read_file`` tool can read it back (issue #3416).
+_VIRTUAL_OUTPUTS_BASE = "/mnt/user-data/outputs"
+
 
 def _default_config() -> ToolOutputConfig:
     return ToolOutputConfig()
@@ -94,6 +105,18 @@ def _sanitize_tool_name(name: str) -> str:
     return safe or "unknown"
 
 
+def _build_externalized_filename(*, tool_name: str, tool_call_id: str) -> str:
+    """Build the on-disk filename for an externalized tool output.
+
+    Shared by the host-disk and sandbox externalization paths so both
+    produce the identical naming scheme.
+    """
+    safe_name = _sanitize_tool_name(tool_name)
+    ext = _EXT_MAP.get(tool_name, "txt")
+    short_id = uuid.uuid4().hex[:12]
+    return f"{safe_name}-{short_id}.{ext}"
+
+
 def _externalize(
     content: str,
     *,
@@ -111,10 +134,7 @@ def _externalize(
     except OSError:
         return None
 
-    safe_name = _sanitize_tool_name(tool_name)
-    ext = _EXT_MAP.get(tool_name, "txt")
-    short_id = uuid.uuid4().hex[:12]
-    filename = f"{safe_name}-{short_id}.{ext}"
+    filename = _build_externalized_filename(tool_name=tool_name, tool_call_id=tool_call_id)
     filepath = os.path.join(storage_dir, filename)
 
     if not os.path.abspath(filepath).startswith(os.path.abspath(storage_dir)):
@@ -126,8 +146,56 @@ def _externalize(
     except OSError:
         return None
 
-    virtual_base = "/mnt/user-data/outputs"
-    return f"{virtual_base}/{storage_subdir}/{filename}"
+    return f"{_VIRTUAL_OUTPUTS_BASE}/{storage_subdir}/{filename}"
+
+
+def _externalize_to_sandbox(
+    content: str,
+    *,
+    tool_name: str,
+    tool_call_id: str,
+    storage_subdir: str,
+    sandbox: Sandbox,
+) -> str | None:
+    """Write *content* into the sandbox filesystem and return the virtual path.
+
+    Used when the sandbox does not use thread-data mounts (e.g. a remote AIO
+    sandbox): the host-side :func:`_externalize` virtual path would not exist
+    inside the sandbox, so the model's ``read_file`` tool could not read it
+    back (issue #3416). Returns the same virtual-path contract on success, or
+    ``None`` to signal the caller to fall back to inline truncation.
+    """
+    if os.path.isabs(storage_subdir) or ".." in storage_subdir:
+        return None
+    filename = _build_externalized_filename(tool_name=tool_name, tool_call_id=tool_call_id)
+    virtual_dir = f"{_VIRTUAL_OUTPUTS_BASE}/{storage_subdir}"
+    virtual_path = f"{virtual_dir}/{filename}"
+    try:
+        # AIO sandbox write_file does NOT create parent directories, so create
+        # them explicitly before writing. execute_command returns its stdout
+        # verbatim (including an "Error: ..." string on failure) rather than
+        # raising, so we cannot rely on exception propagation here.
+        sandbox.execute_command(f"mkdir -p {shlex.quote(virtual_dir)}")
+        sandbox.write_file(virtual_path, content)
+        # Validate the file landed: execute_command may have silently failed
+        # to create the directory, and write_file backends differ. Refuse to
+        # hand the model an unreadable read_file path.
+        check = sandbox.execute_command(f"test -s {shlex.quote(virtual_path)} && echo OK || echo MISSING")
+        if not isinstance(check, str) or check.strip() != "OK":
+            logger.warning(
+                "Sandbox externalize validation failed: path=%s, check=%r",
+                virtual_path,
+                check,
+            )
+            return None
+    except Exception:
+        logger.exception(
+            "Failed to externalize %s output to sandbox (call_id=%s)",
+            tool_name,
+            tool_call_id,
+        )
+        return None
+    return virtual_path
 
 
 # ---------------------------------------------------------------------------
@@ -227,6 +295,33 @@ def _resolve_outputs_path(request: ToolCallRequest) -> str | None:
     return outputs_path if isinstance(outputs_path, str) else None
 
 
+def _resolve_sandbox(request: ToolCallRequest) -> Sandbox | None:
+    """Resolve the active sandbox for the current tool call, or ``None``.
+
+    Reads the sandbox_id that ``SandboxMiddleware`` (and the sandbox tools
+    themselves) write into ``runtime.state["sandbox"]``. We intentionally do
+    NOT call ``provider.acquire`` here: acquiring a sandbox can trigger
+    blocking remote I/O, and this resolver runs on every tool call. Tools
+    that do not use a sandbox (``web_search``, MCP, ...) will return ``None``
+    here, which is fine -- the caller falls back to inline truncation.
+    """
+    runtime = getattr(request, "runtime", None)
+    state = getattr(runtime, "state", None)
+    if not isinstance(state, dict):
+        return None
+    sandbox_state = state.get("sandbox")
+    if not isinstance(sandbox_state, dict):
+        return None
+    sandbox_id = sandbox_state.get("sandbox_id")
+    if not sandbox_id:
+        return None
+    try:
+        return get_sandbox_provider().get(sandbox_id)
+    except Exception:
+        logger.exception("Failed to look up sandbox %s for tool-output externalization", sandbox_id)
+        return None
+
+
 def _budget_content(
     content: str,
     *,
@@ -234,6 +329,7 @@ def _budget_content(
     tool_call_id: str,
     outputs_path: str | None,
     config: ToolOutputConfig,
+    sandbox: Sandbox | None = None,
 ) -> str | None:
     """Apply budget to *content*. Returns ``None`` if no change needed."""
     threshold = config.tool_overrides.get(tool_name, config.externalize_min_chars)
@@ -242,14 +338,50 @@ def _budget_content(
     if len(content) <= threshold and len(content) <= config.fallback_max_chars:
         return None
 
-    if threshold > 0 and len(content) > threshold and outputs_path:
-        virtual_path = _externalize(
-            content,
-            tool_name=tool_name,
-            tool_call_id=tool_call_id,
-            outputs_path=outputs_path,
-            storage_subdir=config.storage_subdir,
-        )
+    if threshold > 0 and len(content) > threshold:
+        virtual_path: str | None = None
+        # Decide persistence target based on what's available, without touching
+        # the sandbox provider unless a sandbox was actually resolved for this
+        # call. This keeps the legacy host-disk path provider-free, so callers
+        # without a configured sandbox (and CI environments without a
+        # config.yaml) continue to externalize to the host as before.
+        if sandbox is not None:
+            provider = None
+            try:
+                provider = get_sandbox_provider()
+            except Exception:
+                logger.exception("Failed to get sandbox provider for tool-output externalization; falling back to inline truncation")
+            if provider is not None and getattr(provider, "uses_thread_data_mounts", False):
+                # Host-mounted sandbox: host outputs path is bind-mounted into
+                # the sandbox at the same virtual path, so writing host-side is
+                # equivalent. Preserve the original behavior to avoid extra
+                # sandbox round-trips.
+                if outputs_path:
+                    virtual_path = _externalize(
+                        content,
+                        tool_name=tool_name,
+                        tool_call_id=tool_call_id,
+                        outputs_path=outputs_path,
+                        storage_subdir=config.storage_subdir,
+                    )
+            else:
+                virtual_path = _externalize_to_sandbox(
+                    content,
+                    tool_name=tool_name,
+                    tool_call_id=tool_call_id,
+                    storage_subdir=config.storage_subdir,
+                    sandbox=sandbox,
+                )
+        elif outputs_path:
+            # No sandbox in this call (legacy / non-sandbox tools): write to
+            # host outputs path directly, no provider needed.
+            virtual_path = _externalize(
+                content,
+                tool_name=tool_name,
+                tool_call_id=tool_call_id,
+                outputs_path=outputs_path,
+                storage_subdir=config.storage_subdir,
+            )
         if virtual_path is not None:
             logger.info(
                 "Externalized %s output (%d chars) to %s",
@@ -288,7 +420,12 @@ def _budget_content(
 # ---------------------------------------------------------------------------
 
 
-def _patch_tool_message(msg: ToolMessage, config: ToolOutputConfig, outputs_path: str | None) -> ToolMessage:
+def _patch_tool_message(
+    msg: ToolMessage,
+    config: ToolOutputConfig,
+    outputs_path: str | None,
+    sandbox: Sandbox | None = None,
+) -> ToolMessage:
     """Apply budget to a single ToolMessage. Returns the original if unchanged."""
     tool_name = msg.name or "unknown"
     if tool_name in config.exempt_tools:
@@ -304,6 +441,7 @@ def _patch_tool_message(msg: ToolMessage, config: ToolOutputConfig, outputs_path
         tool_call_id=msg.tool_call_id or "",
         outputs_path=outputs_path,
         config=config,
+        sandbox=sandbox,
     )
     if replacement is None:
         return msg
@@ -355,10 +493,15 @@ def _needs_budget(result: ToolMessage | Command, config: ToolOutputConfig) -> bo
     return False
 
 
-def _patch_result(result: ToolMessage | Command, config: ToolOutputConfig, outputs_path: str | None) -> ToolMessage | Command:
+def _patch_result(
+    result: ToolMessage | Command,
+    config: ToolOutputConfig,
+    outputs_path: str | None,
+    sandbox: Sandbox | None = None,
+) -> ToolMessage | Command:
     """Apply budget to a tool call result (ToolMessage or Command)."""
     if isinstance(result, ToolMessage):
-        return _patch_tool_message(result, config, outputs_path)
+        return _patch_tool_message(result, config, outputs_path, sandbox)
 
     update = getattr(result, "update", None)
     if not isinstance(update, dict):
@@ -372,7 +515,7 @@ def _patch_result(result: ToolMessage | Command, config: ToolOutputConfig, outpu
     changed = False
     for msg in messages:
         if isinstance(msg, ToolMessage):
-            patched = _patch_tool_message(msg, config, outputs_path)
+            patched = _patch_tool_message(msg, config, outputs_path, sandbox)
             if patched is not msg:
                 changed = True
             new_messages.append(patched)
@@ -392,6 +535,11 @@ def _patch_model_messages(messages: list[Any], config: ToolOutputConfig) -> list
     ToolMessage exceeds the budget — the common case once every result has
     already been budgeted at tool-call time, so a long history is not rebuilt
     on every model call.
+
+    Historical messages do not get a ``sandbox`` argument: any oversized tool
+    message in history was already budgeted (and possibly externalized) at
+    tool-call time, so the only thing left for the history path to do is
+    inline fallback truncation, which needs no sandbox.
     """
     if not any(isinstance(msg, ToolMessage) and _tool_message_over_budget(msg, config) for msg in messages):
         return None
@@ -442,7 +590,8 @@ class ToolOutputBudgetMiddleware(AgentMiddleware[AgentState]):
         if not _needs_budget(result, self._config):
             return result
         outputs_path = _resolve_outputs_path(request)
-        return _patch_result(result, self._config, outputs_path)
+        sandbox = _resolve_sandbox(request)
+        return _patch_result(result, self._config, outputs_path, sandbox)
 
     @override
     async def awrap_tool_call(
@@ -456,7 +605,12 @@ class ToolOutputBudgetMiddleware(AgentMiddleware[AgentState]):
         if not _needs_budget(result, self._config):
             return result
         outputs_path = _resolve_outputs_path(request)
-        return await asyncio.to_thread(_patch_result, result, self._config, outputs_path)
+        # _resolve_sandbox only touches runtime.state and the provider's
+        # in-memory sandbox registry, so it is safe to call on the event
+        # loop. The actual sandbox I/O (mkdir/write/test) happens inside
+        # _patch_result, which is offloaded to a worker thread below.
+        sandbox = _resolve_sandbox(request)
+        return await asyncio.to_thread(_patch_result, result, self._config, outputs_path, sandbox)
 
     # -- model call hooks (historical message truncation) ------------------
 

backend/packages/harness/deerflow/agents/middlewares/uploads_middleware.py

@@ -13,6 +13,7 @@ from langgraph.runtime import Runtime
 from deerflow.config.paths import Paths, get_paths
 from deerflow.runtime.user_context import get_effective_user_id
 from deerflow.utils.file_conversion import extract_outline
+from deerflow.utils.messages import ORIGINAL_USER_CONTENT_KEY, message_content_to_text
 
 logger = logging.getLogger(__name__)
 
@@ -265,6 +266,8 @@ class UploadsMiddleware(AgentMiddleware[UploadsMiddlewareState]):
 
         # Extract original content - handle both string and list formats
         original_content = last_message.content
+        additional_kwargs = dict(last_message.additional_kwargs or {})
+        additional_kwargs.setdefault(ORIGINAL_USER_CONTENT_KEY, message_content_to_text(original_content))
         if isinstance(original_content, str):
             # Simple case: string content, just prepend files message
             updated_content = f"{files_message}\n\n{original_content}"
@@ -285,7 +288,7 @@ class UploadsMiddleware(AgentMiddleware[UploadsMiddlewareState]):
             content=updated_content,
             id=last_message.id,
             name=last_message.name,
-            additional_kwargs=last_message.additional_kwargs,
+            additional_kwargs=additional_kwargs,
         )
 
         messages[last_message_index] = updated_message

backend/packages/harness/deerflow/agents/middlewares/view_image_middleware.py

@@ -179,8 +179,10 @@ class ViewImageMiddleware(AgentMiddleware[ViewImageMiddlewareState]):
         # Create the image details message with text and image content
         image_content = self._create_image_details_message(state)
 
-        # Create a new human message with mixed content (text + images)
-        human_msg = HumanMessage(content=image_content)
+        # Create a new human message with mixed content (text + images). This is
+        # internal context for the model only, so hide it from the chat UI and IM
+        # channels (matches the other middleware-injected context messages).
+        human_msg = HumanMessage(content=image_content, additional_kwargs={"hide_from_ui": True})
 
         logger.debug("Injecting image details message with images before LLM call")
 

backend/packages/harness/deerflow/client.py

@@ -33,7 +33,7 @@ from langchain.agents.middleware import AgentMiddleware
 from langchain_core.messages import AIMessage, HumanMessage, SystemMessage, ToolMessage
 from langchain_core.runnables import RunnableConfig
 
-from deerflow.agents.lead_agent.agent import _assemble_deferred, _build_middlewares
+from deerflow.agents.lead_agent.agent import build_middlewares
 from deerflow.agents.lead_agent.prompt import apply_prompt_template
 from deerflow.agents.thread_state import ThreadState
 from deerflow.config.agents_config import AGENT_NAME_PATTERN
@@ -43,6 +43,7 @@ from deerflow.config.paths import get_paths
 from deerflow.models import create_chat_model
 from deerflow.runtime.user_context import get_effective_user_id
 from deerflow.skills.storage import get_or_new_skill_storage
+from deerflow.tools.builtins.tool_search import assemble_deferred_tools
 from deerflow.tracing import build_tracing_callbacks, inject_langfuse_metadata
 from deerflow.uploads.manager import (
     claim_unique_filename,
@@ -238,7 +239,7 @@ class DeerFlowClient:
         max_concurrent_subagents = cfg.get("max_concurrent_subagents", 3)
 
         tools = self._get_tools(model_name=model_name, subagent_enabled=subagent_enabled)
-        final_tools, deferred_setup = _assemble_deferred(tools, enabled=self._app_config.tool_search.enabled)
+        final_tools, deferred_setup = assemble_deferred_tools(tools, enabled=self._app_config.tool_search.enabled)
         kwargs: dict[str, Any] = {
             # attach_tracing=False because ``stream()`` injects tracing
             # callbacks at the graph invocation root so a single embedded run
@@ -246,7 +247,15 @@ class DeerFlowClient:
             # Attaching them again on the model would emit duplicate spans.
             "model": create_chat_model(name=model_name, thinking_enabled=thinking_enabled, attach_tracing=False),
             "tools": final_tools,
-            "middleware": _build_middlewares(config, model_name=model_name, agent_name=self._agent_name, custom_middlewares=self._middlewares, deferred_setup=deferred_setup),
+            "middleware": build_middlewares(
+                config,
+                model_name=model_name,
+                agent_name=self._agent_name,
+                available_skills=self._available_skills,
+                custom_middlewares=self._middlewares,
+                app_config=self._app_config,
+                deferred_setup=deferred_setup,
+            ),
             "system_prompt": apply_prompt_template(
                 subagent_enabled=subagent_enabled,
                 max_concurrent_subagents=max_concurrent_subagents,

backend/packages/harness/deerflow/community/ddg_search/tools.py

@@ -11,12 +11,85 @@ from deerflow.config import get_app_config
 
 logger = logging.getLogger(__name__)
 
+DEFAULT_BACKEND = "auto"
+DEFAULT_REGION = "wt-wt"
+DEFAULT_SAFESEARCH = "moderate"
+DEFAULT_WIKIPEDIA_REGION = "us-en"
+
+WIKIPEDIA_BACKENDS = {"auto", "all", "wikipedia"}
+WIKIPEDIA_LANGUAGE_ALIASES = {
+    "jp": "ja",
+    "kr": "ko",
+    "tzh": "zh",
+    "wt": "en",
+}
+
+
+def _normalize_backend(backend: str | list[str] | tuple[str, ...] | None) -> str:
+    if backend is None:
+        return DEFAULT_BACKEND
+    if isinstance(backend, (list, tuple)):
+        return ",".join(str(part).strip() for part in backend if str(part).strip()) or DEFAULT_BACKEND
+    return str(backend).strip() or DEFAULT_BACKEND
+
+
+def _normalize_setting(value: str | None, default: str) -> str:
+    return str(value).strip() if value else default
+
+
+def _backend_includes_wikipedia(backend: str | list[str] | tuple[str, ...] | None) -> bool:
+    backend = _normalize_backend(backend)
+    return any(part.strip().lower() in WIKIPEDIA_BACKENDS for part in backend.split(","))
+
+
+def _contains_codepoint(query: str, ranges: tuple[tuple[int, int], ...]) -> bool:
+    return any(start <= ord(char) <= end for char in query for start, end in ranges)
+
+
+def _infer_wikipedia_region(query: str) -> str:
+    """Pick a valid Wikipedia language region when DDGS' worldwide region is used."""
+    if _contains_codepoint(query, ((0x3040, 0x30FF), (0x31F0, 0x31FF))):
+        return "jp-ja"
+    if _contains_codepoint(query, ((0xAC00, 0xD7AF), (0x1100, 0x11FF), (0x3130, 0x318F))):
+        return "kr-ko"
+    if _contains_codepoint(query, ((0x3400, 0x9FFF),)):
+        return "cn-zh"
+    if _contains_codepoint(query, ((0x0400, 0x04FF),)):
+        return "ru-ru"
+    if _contains_codepoint(query, ((0x0370, 0x03FF),)):
+        return "gr-el"
+    if _contains_codepoint(query, ((0x0590, 0x05FF),)):
+        return "il-he"
+    if _contains_codepoint(query, ((0x0600, 0x06FF),)):
+        return "xa-ar"
+    return DEFAULT_WIKIPEDIA_REGION
+
+
+def _resolve_ddgs_region(query: str, region: str | None, backend: str | list[str] | tuple[str, ...] | None) -> str:
+    """
+    DDGS' wikipedia engine treats the second part of region as a Wikipedia
+    subdomain. Its default worldwide region, wt-wt, becomes wt.wikipedia.org.
+    """
+    normalized_region = _normalize_setting(region, DEFAULT_REGION).lower()
+    if not _backend_includes_wikipedia(backend):
+        return normalized_region
+
+    if normalized_region == DEFAULT_REGION:
+        return _infer_wikipedia_region(query)
+
+    if "-" not in normalized_region:
+        return DEFAULT_WIKIPEDIA_REGION
+
+    country, language = normalized_region.split("-", 1)
+    return f"{country}-{WIKIPEDIA_LANGUAGE_ALIASES.get(language, language)}"
+
 
 def _search_text(
     query: str,
     max_results: int = 5,
-    region: str = "wt-wt",
-    safesearch: str = "moderate",
+    region: str | None = DEFAULT_REGION,
+    safesearch: str | None = DEFAULT_SAFESEARCH,
+    backend: str | list[str] | tuple[str, ...] | None = DEFAULT_BACKEND,
 ) -> list[dict]:
     """
     Execute text search using DuckDuckGo.
@@ -26,6 +99,7 @@ def _search_text(
         max_results: Maximum number of results
         region: Search region
         safesearch: Safe search level
+        backend: DDGS backend(s), e.g. "auto", "duckduckgo", or "duckduckgo,brave"
 
     Returns:
         List of search results
@@ -39,11 +113,15 @@ def _search_text(
     ddgs = DDGS(timeout=30)
 
     try:
+        backend = _normalize_backend(backend)
+        safesearch = _normalize_setting(safesearch, DEFAULT_SAFESEARCH)
+        effective_region = _resolve_ddgs_region(query, region, backend)
         results = ddgs.text(
             query,
-            region=region,
+            region=effective_region,
             safesearch=safesearch,
             max_results=max_results,
+            backend=backend,
         )
         return list(results) if results else []
 
@@ -64,14 +142,23 @@ def web_search_tool(
         max_results: Maximum number of results to return. Default is 5.
     """
     config = get_app_config().get_tool_config("web_search")
+    region = DEFAULT_REGION
+    safesearch = DEFAULT_SAFESEARCH
+    backend = DEFAULT_BACKEND
 
-    # Override max_results from config if set
-    if config is not None and "max_results" in config.model_extra:
+    if config is not None:
+        # Override tool call defaults from config if set.
         max_results = config.model_extra.get("max_results", max_results)
+        region = config.model_extra.get("region", region)
+        safesearch = config.model_extra.get("safesearch", safesearch)
+        backend = config.model_extra.get("backend", backend)
 
     results = _search_text(
         query=query,
         max_results=max_results,
+        region=region,
+        safesearch=safesearch,
+        backend=backend,
     )
 
     if not results:

backend/packages/harness/deerflow/community/jina_ai/jina_client.py

@@ -9,7 +9,7 @@ _api_key_warned = False
 
 
 class JinaClient:
-    async def crawl(self, url: str, return_format: str = "html", timeout: int = 10) -> str:
+    async def crawl(self, url: str, return_format: str = "html", timeout: int = 10, proxy: str | None = None, trust_env: bool = True) -> str:
         global _api_key_warned
         headers = {
             "Content-Type": "application/json",
@@ -23,7 +23,10 @@ class JinaClient:
             logger.warning("Jina API key is not set. Provide your own key to access a higher rate limit. See https://jina.ai/reader for more information.")
         data = {"url": url}
         try:
-            async with httpx.AsyncClient() as client:
+            client_kwargs: dict[str, object] = {"trust_env": trust_env}
+            if proxy:
+                client_kwargs["proxy"] = proxy
+            async with httpx.AsyncClient(**client_kwargs) as client:
                 response = await client.post("https://r.jina.ai/", headers=headers, json=data, timeout=timeout)
 
             if response.status_code != 200:

backend/packages/harness/deerflow/community/jina_ai/tools.py

@@ -9,6 +9,38 @@ from deerflow.utils.readability import ReadabilityExtractor
 readability_extractor = ReadabilityExtractor()
 
 
+def _coerce_bool(value: object, default: bool) -> bool:
+    if isinstance(value, bool):
+        return value
+    if isinstance(value, str):
+        normalized = value.strip().lower()
+        if normalized in {"1", "true", "yes", "on"}:
+            return True
+        if normalized in {"0", "false", "no", "off"}:
+            return False
+    return default
+
+
+def _coerce_timeout(value: object, default: int) -> int:
+    if isinstance(value, bool):
+        return default
+    if isinstance(value, int):
+        return value
+    if isinstance(value, str):
+        try:
+            return int(value)
+        except ValueError:
+            return default
+    return default
+
+
+def _coerce_proxy(value: object) -> str | None:
+    if not isinstance(value, str):
+        return None
+    proxy = value.strip()
+    return proxy or None
+
+
 @tool("web_fetch", parse_docstring=True)
 async def web_fetch_tool(url: str) -> str:
     """Fetch the contents of a web page at a given URL.
@@ -22,10 +54,14 @@ async def web_fetch_tool(url: str) -> str:
     """
     jina_client = JinaClient()
     timeout = 10
+    proxy = None
+    trust_env = True
     config = get_app_config().get_tool_config("web_fetch")
-    if config is not None and "timeout" in config.model_extra:
-        timeout = config.model_extra.get("timeout")
-    html_content = await jina_client.crawl(url, return_format="html", timeout=timeout)
+    if config is not None:
+        timeout = _coerce_timeout(config.model_extra.get("timeout"), timeout)
+        proxy = _coerce_proxy(config.model_extra.get("proxy"))
+        trust_env = _coerce_bool(config.model_extra.get("trust_env"), trust_env)
+    html_content = await jina_client.crawl(url, return_format="html", timeout=timeout, proxy=proxy, trust_env=trust_env)
     if isinstance(html_content, str) and html_content.startswith("Error:"):
         return html_content
     article = await asyncio.to_thread(readability_extractor.extract_article, html_content)

backend/packages/harness/deerflow/config/app_config.py

@@ -7,10 +7,11 @@ from typing import Any, Self
 
 import yaml
 from dotenv import load_dotenv
-from pydantic import BaseModel, ConfigDict, Field
+from pydantic import BaseModel, ConfigDict, Field, field_validator
 
 from deerflow.config.acp_config import ACPAgentConfig, load_acp_config_from_dict
 from deerflow.config.agents_api_config import AgentsApiConfig, load_agents_api_config_from_dict
+from deerflow.config.channel_connections_config import ChannelConnectionsConfig
 from deerflow.config.checkpointer_config import CheckpointerConfig, load_checkpointer_config_from_dict
 from deerflow.config.database_config import DatabaseConfig
 from deerflow.config.extensions_config import ExtensionsConfig
@@ -116,6 +117,7 @@ class AppConfig(BaseModel):
     subagents: SubagentsAppConfig = Field(default_factory=SubagentsAppConfig, description="Subagent runtime configuration")
     guardrails: GuardrailsConfig = Field(default_factory=GuardrailsConfig, description="Guardrail middleware configuration")
     circuit_breaker: CircuitBreakerConfig = Field(default_factory=CircuitBreakerConfig, description="LLM circuit breaker configuration")
+    channel_connections: ChannelConnectionsConfig = Field(default_factory=ChannelConnectionsConfig, description="User-facing IM channel connection configuration")
     loop_detection: LoopDetectionConfig = Field(default_factory=LoopDetectionConfig, description="Loop detection middleware configuration")
     safety_finish_reason: SafetyFinishReasonConfig = Field(default_factory=SafetyFinishReasonConfig, description="Provider safety-filter finish_reason interception middleware configuration")
     model_config = ConfigDict(extra="allow")
@@ -148,6 +150,21 @@ class AppConfig(BaseModel):
         ),
     )
 
+    @field_validator("models", "tools", "tool_groups", mode="before")
+    @classmethod
+    def _coerce_null_list_sections(cls, value: Any) -> Any:
+        """Treat a present-but-empty config section as an empty list.
+
+        Commenting out every entry under a top-level YAML key — e.g. ``models:``
+        with only comments beneath it, exactly as shipped in
+        ``config.example.yaml`` — makes PyYAML parse the value as ``None``.
+        Without this, the documented ``cp config.example.yaml config.yaml``
+        first-run flow crashes with an opaque ``Input should be a valid list``
+        pydantic error. Coercing ``None`` to ``[]`` keeps that flow working and
+        matches the field's own ``default_factory=list``.
+        """
+        return [] if value is None else value
+
     @classmethod
     def resolve_config_path(cls, config_path: str | None = None) -> Path:
         """Resolve the config file path.
@@ -209,6 +226,11 @@ class AppConfig(BaseModel):
         config_data["extensions"] = extensions_config.model_dump()
 
         result = cls.model_validate(config_data)
+        if not result.models:
+            logger.warning(
+                "No models are configured in %s. Add at least one entry under `models:` (see the commented examples in config.example.yaml) or run `make setup`.",
+                resolved_path,
+            )
         acp_agents = cls._validate_acp_agents(config_data.get("acp_agents", {}))
         cls._apply_singleton_configs(result, acp_agents)
         return result

backend/packages/harness/deerflow/config/channel_connections_config.py

@@ -0,0 +1,49 @@
+"""Configuration for user-owned IM channel connections."""
+
+from __future__ import annotations
+
+from pydantic import BaseModel, Field
+
+
+class SlackChannelConnectionConfig(BaseModel):
+    enabled: bool = False
+
+    @property
+    def configured(self) -> bool:
+        return True
+
+
+class TelegramChannelConnectionConfig(BaseModel):
+    enabled: bool = False
+    bot_username: str = ""
+
+    @property
+    def configured(self) -> bool:
+        return bool(self.bot_username)
+
+
+class DiscordChannelConnectionConfig(BaseModel):
+    enabled: bool = False
+
+    @property
+    def configured(self) -> bool:
+        return True
+
+
+class ChannelConnectionsConfig(BaseModel):
+    """Top-level config for browser-connectable IM channels."""
+
+    enabled: bool = False
+    slack: SlackChannelConnectionConfig = Field(default_factory=SlackChannelConnectionConfig)
+    telegram: TelegramChannelConnectionConfig = Field(default_factory=TelegramChannelConnectionConfig)
+    discord: DiscordChannelConnectionConfig = Field(default_factory=DiscordChannelConnectionConfig)
+
+    def provider_status(self, provider: str) -> dict[str, bool]:
+        config = getattr(self, provider, None)
+        if config is None:
+            return {"enabled": False, "configured": False}
+        enabled = bool(config.enabled)
+        return {
+            "enabled": enabled,
+            "configured": enabled and bool(config.configured),
+        }

backend/packages/harness/deerflow/config/checkpointer_config.py

@@ -41,6 +41,20 @@ def set_checkpointer_config(config: CheckpointerConfig | None) -> None:
     _checkpointer_config = config
 
 
+def ensure_config_loaded() -> None:
+    """Lazily load app config when checkpointer config has not been initialized."""
+    from deerflow.config.app_config import _app_config, get_app_config
+
+    config = get_checkpointer_config()
+    if config is not None or _app_config is not None:
+        return
+
+    try:
+        get_app_config()
+    except FileNotFoundError:
+        pass
+
+
 def load_checkpointer_config_from_dict(config_dict: dict | None) -> None:
     """Load checkpointer configuration from a dictionary."""
     global _checkpointer_config

backend/packages/harness/deerflow/config/paths.py

@@ -47,7 +47,7 @@ def make_safe_user_id(raw: str) -> str:
     sanitized = _UNSAFE_USER_ID_CHAR_RE.sub("-", raw)
     if sanitized == raw:
         return raw
-    digest = hashlib.sha1(raw.encode("utf-8")).hexdigest()[:_SAFE_USER_ID_DIGEST_HEX_LEN]
+    digest = hashlib.sha256(raw.encode("utf-8")).hexdigest()[:_SAFE_USER_ID_DIGEST_HEX_LEN]
     return f"{sanitized}-{digest}"
 
 

backend/packages/harness/deerflow/config/sandbox_config.py

@@ -4,7 +4,20 @@ from pydantic import BaseModel, ConfigDict, Field
 class VolumeMountConfig(BaseModel):
     """Configuration for a volume mount."""
 
-    host_path: str = Field(..., description="Path on the host machine")
+    host_path: str = Field(
+        ...,
+        description=(
+            "Source path for the mount. Resolution depends on the active provider: "
+            "``LocalSandboxProvider`` checks this path from the gateway process — in "
+            "``make dev`` that is the host machine, but in Docker deployments "
+            "(``make up`` / docker-compose) it is the path *inside* the "
+            "``deer-flow-gateway`` container, so the host directory must also be "
+            "bind-mounted into the gateway service for the mount to take effect. "
+            "``AioSandboxProvider`` (DooD) passes this value straight to ``docker -v`` "
+            "for the sandbox container, where it is resolved by the host Docker daemon "
+            "from the host machine's perspective."
+        ),
+    )
     container_path: str = Field(..., description="Path inside the container")
     read_only: bool = Field(default=False, description="Whether the mount is read-only")
 

backend/packages/harness/deerflow/models/patched_minimax.py

@@ -114,8 +114,27 @@ class PatchedChatMiniMax(ChatOpenAI):
             }
         else:
             payload["extra_body"] = {"reasoning_split": True}
+        self._strip_user_message_names(payload)
         return payload
 
+    @staticmethod
+    def _strip_user_message_names(payload: dict) -> None:
+        """Drop the per-message ``name`` field from user-role messages.
+
+        DeerFlow middlewares tag user messages with internal provenance names
+        (``user-input``, ``summary``, ``loop_warning``, ...). ``langchain_openai``
+        serializes those into the OpenAI-compatible request, but MiniMax requires
+        every user-role ``name`` to be identical and otherwise rejects the request
+        with ``invalid params, user name must be consistent (2013)``. MiniMax does
+        not use the per-message author name, so strip it.
+        """
+        messages = payload.get("messages")
+        if not isinstance(messages, list):
+            return
+        for message in messages:
+            if isinstance(message, dict) and message.get("role") == "user":
+                message.pop("name", None)
+
     def _convert_chunk_to_generation_chunk(
         self,
         chunk: dict,

backend/packages/harness/deerflow/models/patched_stepfun.py

@@ -0,0 +1,175 @@
+"""Patched ChatOpenAI adapter for StepFun reasoning models.
+
+StepFun returns ``reasoning`` (or ``reasoning_content`` with deepseek-style) in
+both streaming deltas and non-streaming responses. Standard ``ChatOpenAI``
+ignores these non-standard fields, so reasoning content is silently dropped.
+This adapter captures reasoning from all response paths and replays it on
+historical assistant messages for multi-turn tool-call conversations.
+"""
+
+from __future__ import annotations
+
+from collections.abc import Mapping
+from typing import Any
+
+from langchain_core.language_models import LanguageModelInput
+from langchain_core.messages import AIMessage, AIMessageChunk
+from langchain_core.outputs import ChatGeneration, ChatGenerationChunk, ChatResult
+from langchain_openai import ChatOpenAI
+
+from deerflow.models.assistant_payload_replay import (
+    restore_assistant_payloads,
+    restore_reasoning_content,
+)
+
+_MISSING = object()
+
+
+def _extract_reasoning(value: Any) -> str | object:
+    """Return reasoning content from a dict/Pydantic object.
+
+    StepFun may return reasoning via ``reasoning`` (default) or
+    ``reasoning_content`` (deepseek-style). Check both fields.
+    """
+    if isinstance(value, Mapping):
+        # Check reasoning_content first (deepseek-style), then reasoning (default)
+        for field in ("reasoning_content", "reasoning"):
+            if field in value and value[field] is not None:
+                return value[field]
+        return _MISSING
+
+    # Pydantic / SDK object attributes
+    for field in ("reasoning_content", "reasoning"):
+        attr = getattr(value, field, _MISSING)
+        if attr is not _MISSING and attr is not None:
+            return attr
+
+    # Some SDK versions store extra fields in model_extra
+    model_extra = getattr(value, "model_extra", None)
+    if isinstance(model_extra, Mapping):
+        for field in ("reasoning_content", "reasoning"):
+            if field in model_extra and model_extra[field] is not None:
+                return model_extra[field]
+
+    return _MISSING
+
+
+def _with_reasoning_content(message: AIMessage | AIMessageChunk, reasoning: str) -> AIMessage | AIMessageChunk:
+    """Return a copy of *message* with reasoning_content stored in additional_kwargs."""
+    additional_kwargs = dict(message.additional_kwargs)
+    if additional_kwargs.get("reasoning_content") != reasoning:
+        additional_kwargs["reasoning_content"] = reasoning
+    return message.model_copy(update={"additional_kwargs": additional_kwargs})
+
+
+def _get_typed_choice_message(response: Any, index: int) -> Any:
+    """Extract the SDK-typed choice message at *index*, if available."""
+    choices = getattr(response, "choices", None)
+    if choices is None:
+        return None
+    try:
+        return choices[index].message
+    except (AttributeError, IndexError, TypeError):
+        return None
+
+
+class PatchedChatStepFun(ChatOpenAI):
+    """ChatOpenAI with full reasoning support for StepFun models.
+
+    Captures ``reasoning`` / ``reasoning_content`` from both streaming and
+    non-streaming responses and replays it on historical assistant messages in
+    multi-turn tool-call conversations.
+    """
+
+    @classmethod
+    def is_lc_serializable(cls) -> bool:
+        return True
+
+    @property
+    def lc_secrets(self) -> dict[str, str]:
+        return {"api_key": "STEPFUN_API_KEY", "openai_api_key": "STEPFUN_API_KEY"}
+
+    # --- Request payload replay ---
+
+    def _get_request_payload(
+        self,
+        input_: LanguageModelInput,
+        *,
+        stop: list[str] | None = None,
+        **kwargs: Any,
+    ) -> dict:
+        """Restore ``reasoning_content`` on historical assistant messages."""
+        original_messages = self._convert_input(input_).to_messages()
+        payload = super()._get_request_payload(input_, stop=stop, **kwargs)
+
+        restore_assistant_payloads(
+            payload.get("messages", []),
+            original_messages,
+            restore_reasoning_content,
+        )
+
+        return payload
+
+    # --- Streaming reasoning capture ---
+
+    def _convert_chunk_to_generation_chunk(
+        self,
+        chunk: dict,
+        default_chunk_class: type,
+        base_generation_info: dict | None,
+    ) -> ChatGenerationChunk | None:
+        """Capture ``reasoning`` / ``reasoning_content`` from streaming deltas."""
+        generation_chunk = super()._convert_chunk_to_generation_chunk(
+            chunk,
+            default_chunk_class,
+            base_generation_info,
+        )
+        if generation_chunk is None:
+            return None
+
+        choices = chunk.get("choices", [])
+        if choices:
+            delta = choices[0].get("delta") or {}
+            reasoning = _extract_reasoning(delta)
+            if reasoning is not _MISSING and isinstance(generation_chunk.message, AIMessageChunk):
+                generation_chunk = ChatGenerationChunk(
+                    message=_with_reasoning_content(generation_chunk.message, reasoning),
+                    generation_info=generation_chunk.generation_info,
+                )
+
+        return generation_chunk
+
+    # --- Non-streaming reasoning capture ---
+
+    def _create_chat_result(
+        self,
+        response: dict | Any,
+        generation_info: dict | None = None,
+    ) -> ChatResult:
+        """Extract ``reasoning`` / ``reasoning_content`` from non-streaming responses."""
+        result = super()._create_chat_result(response, generation_info)
+        response_dict = response if isinstance(response, dict) else response.model_dump()
+        choices = response_dict.get("choices", [])
+
+        patched_generations: list[ChatGeneration] | None = None
+        for index, generation in enumerate(result.generations):
+            choice = choices[index] if index < len(choices) else {}
+            choice_message = choice.get("message", {}) if isinstance(choice, Mapping) else {}
+            reasoning = _extract_reasoning(choice_message)
+
+            if reasoning is _MISSING and not isinstance(response, dict):
+                reasoning = _extract_reasoning(_get_typed_choice_message(response, index))
+
+            message = generation.message
+            if reasoning is not _MISSING and isinstance(message, AIMessage):
+                if patched_generations is None:
+                    patched_generations = list(result.generations)
+                patched_generations[index] = ChatGeneration(
+                    message=_with_reasoning_content(message, reasoning),
+                    generation_info=generation.generation_info,
+                )
+
+        return ChatResult(
+            generations=patched_generations or result.generations,
+            llm_output=result.llm_output,
+        )

backend/packages/harness/deerflow/persistence/channel_connections/__init__.py

@@ -0,0 +1,21 @@
+"""User-owned IM channel connection persistence."""
+
+from deerflow.persistence.channel_connections.model import (
+    ChannelConnectionRow,
+    ChannelConversationRow,
+    ChannelCredentialRow,
+    ChannelOAuthStateRow,
+)
+from deerflow.persistence.channel_connections.sql import (
+    ChannelConnectionRepository,
+    ChannelCredentialCipher,
+)
+
+__all__ = [
+    "ChannelConnectionRepository",
+    "ChannelConnectionRow",
+    "ChannelConversationRow",
+    "ChannelCredentialCipher",
+    "ChannelCredentialRow",
+    "ChannelOAuthStateRow",
+]

backend/packages/harness/deerflow/persistence/channel_connections/model.py

@@ -0,0 +1,111 @@
+"""ORM models for user-owned IM channel connections."""
+
+from __future__ import annotations
+
+from datetime import UTC, datetime
+
+from sqlalchemy import JSON, DateTime, ForeignKey, Index, Integer, String, Text, UniqueConstraint
+from sqlalchemy.orm import Mapped, mapped_column
+
+from deerflow.persistence.base import Base
+
+
+def _utc_now() -> datetime:
+    return datetime.now(UTC)
+
+
+class ChannelConnectionRow(Base):
+    __tablename__ = "channel_connections"
+
+    id: Mapped[str] = mapped_column(String(64), primary_key=True)
+    owner_user_id: Mapped[str] = mapped_column(String(64), nullable=False, index=True)
+    provider: Mapped[str] = mapped_column(String(32), nullable=False, index=True)
+    status: Mapped[str] = mapped_column(String(32), nullable=False, default="connected")
+
+    external_account_id: Mapped[str] = mapped_column(String(128), nullable=False, default="")
+    external_account_name: Mapped[str | None] = mapped_column(String(256), nullable=True)
+    workspace_id: Mapped[str] = mapped_column(String(128), nullable=False, default="")
+    workspace_name: Mapped[str | None] = mapped_column(String(256), nullable=True)
+    bot_user_id: Mapped[str | None] = mapped_column(String(128), nullable=True)
+
+    scopes_json: Mapped[list] = mapped_column(JSON, default=list)
+    capabilities_json: Mapped[dict] = mapped_column(JSON, default=dict)
+    metadata_json: Mapped[dict] = mapped_column(JSON, default=dict)
+
+    created_at: Mapped[datetime] = mapped_column(DateTime(timezone=True), nullable=False, default=_utc_now)
+    updated_at: Mapped[datetime] = mapped_column(DateTime(timezone=True), nullable=False, default=_utc_now, onupdate=_utc_now)
+    last_seen_at: Mapped[datetime | None] = mapped_column(DateTime(timezone=True), nullable=True)
+    last_error_at: Mapped[datetime | None] = mapped_column(DateTime(timezone=True), nullable=True)
+
+    __table_args__ = (
+        UniqueConstraint(
+            "owner_user_id",
+            "provider",
+            "external_account_id",
+            "workspace_id",
+            name="uq_channel_connection_owner_provider_identity",
+        ),
+        Index("idx_channel_connections_event_lookup", "provider", "workspace_id", "bot_user_id"),
+    )
+
+
+class ChannelCredentialRow(Base):
+    __tablename__ = "channel_credentials"
+
+    connection_id: Mapped[str] = mapped_column(
+        String(64),
+        ForeignKey("channel_connections.id", ondelete="CASCADE"),
+        primary_key=True,
+    )
+    encrypted_access_token: Mapped[str | None] = mapped_column(Text, nullable=True)
+    encrypted_refresh_token: Mapped[str | None] = mapped_column(Text, nullable=True)
+    token_type: Mapped[str | None] = mapped_column(String(32), nullable=True)
+    expires_at: Mapped[datetime | None] = mapped_column(DateTime(timezone=True), nullable=True)
+    refresh_expires_at: Mapped[datetime | None] = mapped_column(DateTime(timezone=True), nullable=True)
+    encrypted_extra_json: Mapped[str | None] = mapped_column(Text, nullable=True)
+    version: Mapped[int] = mapped_column(Integer, nullable=False, default=1)
+    updated_at: Mapped[datetime] = mapped_column(DateTime(timezone=True), nullable=False, default=_utc_now, onupdate=_utc_now)
+
+
+class ChannelOAuthStateRow(Base):
+    __tablename__ = "channel_oauth_states"
+
+    state_hash: Mapped[str] = mapped_column(String(128), primary_key=True)
+    owner_user_id: Mapped[str] = mapped_column(String(64), nullable=False, index=True)
+    provider: Mapped[str] = mapped_column(String(32), nullable=False, index=True)
+    code_verifier_encrypted: Mapped[str | None] = mapped_column(Text, nullable=True)
+    nonce_hash: Mapped[str | None] = mapped_column(String(128), nullable=True)
+    redirect_after: Mapped[str | None] = mapped_column(Text, nullable=True)
+    requested_scopes_json: Mapped[list] = mapped_column(JSON, default=list)
+    metadata_json: Mapped[dict] = mapped_column(JSON, default=dict)
+    expires_at: Mapped[datetime] = mapped_column(DateTime(timezone=True), nullable=False)
+    consumed_at: Mapped[datetime | None] = mapped_column(DateTime(timezone=True), nullable=True)
+    created_at: Mapped[datetime] = mapped_column(DateTime(timezone=True), nullable=False, default=_utc_now)
+
+
+class ChannelConversationRow(Base):
+    __tablename__ = "channel_conversations"
+
+    id: Mapped[str] = mapped_column(String(64), primary_key=True)
+    connection_id: Mapped[str] = mapped_column(
+        String(64),
+        ForeignKey("channel_connections.id", ondelete="CASCADE"),
+        nullable=False,
+        index=True,
+    )
+    owner_user_id: Mapped[str] = mapped_column(String(64), nullable=False, index=True)
+    provider: Mapped[str] = mapped_column(String(32), nullable=False, index=True)
+    external_conversation_id: Mapped[str] = mapped_column(String(128), nullable=False)
+    external_topic_id: Mapped[str] = mapped_column(String(128), nullable=False, default="")
+    thread_id: Mapped[str] = mapped_column(String(64), nullable=False, index=True)
+    created_at: Mapped[datetime] = mapped_column(DateTime(timezone=True), nullable=False, default=_utc_now)
+    updated_at: Mapped[datetime] = mapped_column(DateTime(timezone=True), nullable=False, default=_utc_now, onupdate=_utc_now)
+
+    __table_args__ = (
+        UniqueConstraint(
+            "connection_id",
+            "external_conversation_id",
+            "external_topic_id",
+            name="uq_channel_conversation_connection_external",
+        ),
+    )

backend/packages/harness/deerflow/persistence/channel_connections/sql.py

@@ -0,0 +1,346 @@
+"""SQL repository for user-owned IM channel connections."""
+
+from __future__ import annotations
+
+import base64
+import hashlib
+import json
+import uuid
+from datetime import UTC, datetime
+from typing import Any
+
+from cryptography.fernet import Fernet
+from sqlalchemy import select
+from sqlalchemy.ext.asyncio import AsyncSession, async_sessionmaker
+
+from deerflow.persistence.channel_connections.model import (
+    ChannelConnectionRow,
+    ChannelConversationRow,
+    ChannelCredentialRow,
+    ChannelOAuthStateRow,
+)
+from deerflow.utils.time import coerce_iso
+
+
+class ChannelCredentialCipher:
+    """Encrypts provider credentials before they are persisted."""
+
+    def __init__(self, fernet: Fernet) -> None:
+        self._fernet = fernet
+
+    @classmethod
+    def from_key(cls, key: str) -> ChannelCredentialCipher:
+        digest = hashlib.sha256(key.encode("utf-8")).digest()
+        return cls(Fernet(base64.urlsafe_b64encode(digest)))
+
+    def encrypt_text(self, value: str | None) -> str | None:
+        if value is None:
+            return None
+        return "fernet:v1:" + self._fernet.encrypt(value.encode("utf-8")).decode("ascii")
+
+    def decrypt_text(self, value: str | None) -> str | None:
+        if value is None:
+            return None
+        token = value.removeprefix("fernet:v1:")
+        return self._fernet.decrypt(token.encode("ascii")).decode("utf-8")
+
+
+class ChannelConnectionRepository:
+    """Persistence facade for channel connections, credentials, and conversations."""
+
+    def __init__(
+        self,
+        session_factory: async_sessionmaker[AsyncSession],
+        *,
+        cipher: ChannelCredentialCipher | None = None,
+    ) -> None:
+        self.session_factory = session_factory
+        self._cipher = cipher
+
+    async def close(self) -> None:
+        from deerflow.persistence.engine import close_engine
+
+        await close_engine()
+
+    @staticmethod
+    def _new_id() -> str:
+        return uuid.uuid4().hex
+
+    @staticmethod
+    def _normalize_optional_identity(value: str | None) -> str:
+        return value or ""
+
+    @staticmethod
+    def _coerce_datetime(value: datetime | None) -> datetime | None:
+        if value is None or value.tzinfo is not None:
+            return value
+        return value.replace(tzinfo=UTC)
+
+    def _encrypt_optional_secret(self, value: str | None) -> str | None:
+        if value is None:
+            return None
+        if self._cipher is None:
+            raise RuntimeError("channel connection encryption key is required")
+        return self._cipher.encrypt_text(value)
+
+    @staticmethod
+    def _connection_to_dict(row: ChannelConnectionRow) -> dict[str, Any]:
+        data = row.to_dict()
+        data["external_account_id"] = data["external_account_id"] or None
+        data["workspace_id"] = data["workspace_id"] or None
+        data["scopes"] = data.pop("scopes_json") or []
+        data["capabilities"] = data.pop("capabilities_json") or {}
+        data["metadata"] = data.pop("metadata_json") or {}
+        for key in ("created_at", "updated_at", "last_seen_at", "last_error_at"):
+            value = data.get(key)
+            if isinstance(value, datetime):
+                data[key] = coerce_iso(value)
+        return data
+
+    async def upsert_connection(
+        self,
+        *,
+        owner_user_id: str,
+        provider: str,
+        external_account_id: str | None = None,
+        external_account_name: str | None = None,
+        workspace_id: str | None = None,
+        workspace_name: str | None = None,
+        bot_user_id: str | None = None,
+        scopes: list[str] | None = None,
+        capabilities: dict[str, Any] | None = None,
+        metadata: dict[str, Any] | None = None,
+        status: str = "connected",
+    ) -> dict[str, Any]:
+        external_account_id_value = self._normalize_optional_identity(external_account_id)
+        workspace_id_value = self._normalize_optional_identity(workspace_id)
+        async with self.session_factory() as session:
+            stmt = select(ChannelConnectionRow).where(
+                ChannelConnectionRow.owner_user_id == owner_user_id,
+                ChannelConnectionRow.provider == provider,
+                ChannelConnectionRow.external_account_id == external_account_id_value,
+                ChannelConnectionRow.workspace_id == workspace_id_value,
+            )
+            row = (await session.execute(stmt)).scalar_one_or_none()
+            if row is None:
+                row = ChannelConnectionRow(
+                    id=self._new_id(),
+                    owner_user_id=owner_user_id,
+                    provider=provider,
+                    external_account_id=external_account_id_value,
+                    workspace_id=workspace_id_value,
+                )
+                session.add(row)
+
+            row.status = status
+            row.external_account_name = external_account_name
+            row.workspace_name = workspace_name
+            row.bot_user_id = bot_user_id
+            row.scopes_json = list(scopes or [])
+            row.capabilities_json = dict(capabilities or {})
+            row.metadata_json = dict(metadata or {})
+            await session.commit()
+            await session.refresh(row)
+            return self._connection_to_dict(row)
+
+    async def list_connections(self, owner_user_id: str) -> list[dict[str, Any]]:
+        async with self.session_factory() as session:
+            result = await session.execute(select(ChannelConnectionRow).where(ChannelConnectionRow.owner_user_id == owner_user_id).order_by(ChannelConnectionRow.updated_at.desc(), ChannelConnectionRow.id.desc()))
+            return [self._connection_to_dict(row) for row in result.scalars()]
+
+    async def disconnect_connection(self, *, connection_id: str, owner_user_id: str) -> bool:
+        async with self.session_factory() as session:
+            row = await session.get(ChannelConnectionRow, connection_id)
+            if row is None or row.owner_user_id != owner_user_id:
+                return False
+
+            row.status = "revoked"
+            credential = await session.get(ChannelCredentialRow, connection_id)
+            if credential is not None:
+                await session.delete(credential)
+            await session.commit()
+            return True
+
+    async def store_credentials(
+        self,
+        connection_id: str,
+        *,
+        access_token: str | None,
+        refresh_token: str | None = None,
+        token_type: str | None = None,
+        expires_at: datetime | None = None,
+        refresh_expires_at: datetime | None = None,
+        extra: dict[str, Any] | None = None,
+    ) -> None:
+        if self._cipher is None:
+            raise RuntimeError("channel connection encryption key is required")
+        async with self.session_factory() as session:
+            row = await session.get(ChannelCredentialRow, connection_id)
+            if row is None:
+                row = ChannelCredentialRow(connection_id=connection_id)
+                session.add(row)
+            row.encrypted_access_token = self._cipher.encrypt_text(access_token)
+            row.encrypted_refresh_token = self._cipher.encrypt_text(refresh_token)
+            row.token_type = token_type
+            row.expires_at = expires_at
+            row.refresh_expires_at = refresh_expires_at
+            row.encrypted_extra_json = self._cipher.encrypt_text(json.dumps(extra or {}, ensure_ascii=False))
+            row.version = (row.version or 0) + 1
+            await session.commit()
+
+    async def get_credentials(self, connection_id: str) -> dict[str, Any] | None:
+        if self._cipher is None:
+            return None
+        async with self.session_factory() as session:
+            row = await session.get(ChannelCredentialRow, connection_id)
+            if row is None:
+                return None
+            extra_raw = self._cipher.decrypt_text(row.encrypted_extra_json)
+            return {
+                "connection_id": row.connection_id,
+                "access_token": self._cipher.decrypt_text(row.encrypted_access_token),
+                "refresh_token": self._cipher.decrypt_text(row.encrypted_refresh_token),
+                "token_type": row.token_type,
+                "expires_at": self._coerce_datetime(row.expires_at),
+                "refresh_expires_at": self._coerce_datetime(row.refresh_expires_at),
+                "extra": json.loads(extra_raw) if extra_raw else {},
+            }
+
+    @staticmethod
+    def hash_state(state: str) -> str:
+        return hashlib.sha256(state.encode("utf-8")).hexdigest()
+
+    async def create_oauth_state(
+        self,
+        *,
+        owner_user_id: str,
+        provider: str,
+        state: str,
+        expires_at: datetime,
+        code_verifier: str | None = None,
+        nonce_hash: str | None = None,
+        redirect_after: str | None = None,
+        requested_scopes: list[str] | None = None,
+        metadata: dict[str, Any] | None = None,
+    ) -> None:
+        row = ChannelOAuthStateRow(
+            state_hash=self.hash_state(state),
+            owner_user_id=owner_user_id,
+            provider=provider,
+            code_verifier_encrypted=self._encrypt_optional_secret(code_verifier),
+            nonce_hash=nonce_hash,
+            redirect_after=redirect_after,
+            requested_scopes_json=list(requested_scopes or []),
+            metadata_json=dict(metadata or {}),
+            expires_at=expires_at,
+        )
+        async with self.session_factory() as session:
+            session.add(row)
+            await session.commit()
+
+    async def count_oauth_states(self, *, owner_user_id: str, provider: str) -> int:
+        async with self.session_factory() as session:
+            result = await session.execute(
+                select(ChannelOAuthStateRow).where(
+                    ChannelOAuthStateRow.owner_user_id == owner_user_id,
+                    ChannelOAuthStateRow.provider == provider,
+                )
+            )
+            return len(list(result.scalars()))
+
+    async def consume_oauth_state(
+        self,
+        *,
+        provider: str,
+        state: str,
+        now: datetime | None = None,
+    ) -> dict[str, Any] | None:
+        current_time = now or datetime.now(UTC)
+        async with self.session_factory() as session:
+            row = await session.get(ChannelOAuthStateRow, self.hash_state(state))
+            if row is None or row.provider != provider or row.consumed_at is not None:
+                return None
+            expires_at = self._coerce_datetime(row.expires_at)
+            if expires_at is not None and expires_at < current_time:
+                return None
+
+            row.consumed_at = current_time
+            await session.commit()
+            return {
+                "owner_user_id": row.owner_user_id,
+                "provider": row.provider,
+                "requested_scopes": row.requested_scopes_json or [],
+                "metadata": row.metadata_json or {},
+                "redirect_after": row.redirect_after,
+            }
+
+    async def find_connection_by_external_identity(
+        self,
+        *,
+        provider: str,
+        external_account_id: str,
+        workspace_id: str | None = None,
+    ) -> dict[str, Any] | None:
+        async with self.session_factory() as session:
+            result = await session.execute(
+                select(ChannelConnectionRow)
+                .where(
+                    ChannelConnectionRow.provider == provider,
+                    ChannelConnectionRow.external_account_id == self._normalize_optional_identity(external_account_id),
+                    ChannelConnectionRow.workspace_id == self._normalize_optional_identity(workspace_id),
+                    ChannelConnectionRow.status == "connected",
+                )
+                .order_by(ChannelConnectionRow.updated_at.desc(), ChannelConnectionRow.id.desc())
+                .limit(1)
+            )
+            row = result.scalar_one_or_none()
+            return self._connection_to_dict(row) if row is not None else None
+
+    async def set_thread_id(
+        self,
+        *,
+        connection_id: str,
+        owner_user_id: str,
+        provider: str,
+        external_conversation_id: str,
+        thread_id: str,
+        external_topic_id: str | None = None,
+    ) -> None:
+        topic_id = external_topic_id or ""
+        async with self.session_factory() as session:
+            stmt = select(ChannelConversationRow).where(
+                ChannelConversationRow.connection_id == connection_id,
+                ChannelConversationRow.external_conversation_id == external_conversation_id,
+                ChannelConversationRow.external_topic_id == topic_id,
+            )
+            row = (await session.execute(stmt)).scalar_one_or_none()
+            if row is None:
+                row = ChannelConversationRow(
+                    id=self._new_id(),
+                    connection_id=connection_id,
+                    owner_user_id=owner_user_id,
+                    provider=provider,
+                    external_conversation_id=external_conversation_id,
+                    external_topic_id=topic_id,
+                    thread_id=thread_id,
+                )
+                session.add(row)
+            else:
+                row.thread_id = thread_id
+                row.owner_user_id = owner_user_id
+                row.provider = provider
+            await session.commit()
+
+    async def get_thread_id(
+        self,
+        connection_id: str,
+        external_conversation_id: str,
+        external_topic_id: str | None = None,
+    ) -> str | None:
+        async with self.session_factory() as session:
+            stmt = select(ChannelConversationRow.thread_id).where(
+                ChannelConversationRow.connection_id == connection_id,
+                ChannelConversationRow.external_conversation_id == external_conversation_id,
+                ChannelConversationRow.external_topic_id == (external_topic_id or ""),
+            )
+            return (await session.execute(stmt)).scalar_one_or_none()

backend/packages/harness/deerflow/persistence/models/__init__.py

@@ -14,10 +14,26 @@ its storage implementation lives in ``deerflow.runtime.events.store.db`` and
 there is no matching entity directory.
 """
 
+from deerflow.persistence.channel_connections.model import (
+    ChannelConnectionRow,
+    ChannelConversationRow,
+    ChannelCredentialRow,
+    ChannelOAuthStateRow,
+)
 from deerflow.persistence.feedback.model import FeedbackRow
 from deerflow.persistence.models.run_event import RunEventRow
 from deerflow.persistence.run.model import RunRow
 from deerflow.persistence.thread_meta.model import ThreadMetaRow
 from deerflow.persistence.user.model import UserRow
 
-__all__ = ["FeedbackRow", "RunEventRow", "RunRow", "ThreadMetaRow", "UserRow"]
+__all__ = [
+    "ChannelConnectionRow",
+    "ChannelConversationRow",
+    "ChannelCredentialRow",
+    "ChannelOAuthStateRow",
+    "FeedbackRow",
+    "RunEventRow",
+    "RunRow",
+    "ThreadMetaRow",
+    "UserRow",
+]

backend/packages/harness/deerflow/runtime/checkpointer/provider.py

@@ -21,12 +21,13 @@ from __future__ import annotations
 
 import contextlib
 import logging
+import threading
 from collections.abc import Iterator
 
 from langgraph.types import Checkpointer
 
 from deerflow.config.app_config import get_app_config
-from deerflow.config.checkpointer_config import CheckpointerConfig
+from deerflow.config.checkpointer_config import CheckpointerConfig, ensure_config_loaded
 from deerflow.runtime.store._sqlite_utils import ensure_sqlite_parent_dir, resolve_sqlite_conn_str
 
 logger = logging.getLogger(__name__)
@@ -100,6 +101,7 @@ def _sync_checkpointer_cm(config: CheckpointerConfig) -> Iterator[Checkpointer]:
 
 _checkpointer: Checkpointer | None = None
 _checkpointer_ctx = None  # open context manager keeping the connection alive
+_checkpointer_lock = threading.Lock()
 
 
 def get_checkpointer() -> Checkpointer:
@@ -116,34 +118,29 @@ def get_checkpointer() -> Checkpointer:
     if _checkpointer is not None:
         return _checkpointer
 
-    # Ensure app config is loaded before checking checkpointer config
-    # This prevents returning InMemorySaver when config.yaml actually has a checkpointer section
-    # but hasn't been loaded yet
-    from deerflow.config.app_config import _app_config
-    from deerflow.config.checkpointer_config import get_checkpointer_config
+    # Config loading can reset both persistence singletons. Keep it outside
+    # this provider lock to avoid cross-provider lock-order inversion.
+    ensure_config_loaded()
 
-    config = get_checkpointer_config()
+    with _checkpointer_lock:
+        if _checkpointer is not None:
+            return _checkpointer
+
+        from deerflow.config.checkpointer_config import get_checkpointer_config
 
-    if config is None and _app_config is None:
-        # Only load app config lazily when neither the app config nor an explicit
-        # checkpointer config has been initialized yet. This keeps tests that
-        # intentionally set the global checkpointer config isolated from any
-        # ambient config.yaml on disk.
-        try:
-            get_app_config()
-        except FileNotFoundError:
-            # In test environments without config.yaml, this is expected.
-            pass
         config = get_checkpointer_config()
-    if config is None:
-        from langgraph.checkpoint.memory import InMemorySaver
 
-        logger.info("Checkpointer: using InMemorySaver (in-process, not persistent)")
-        _checkpointer = InMemorySaver()
-        return _checkpointer
+        if config is None:
+            from langgraph.checkpoint.memory import InMemorySaver
 
-    _checkpointer_ctx = _sync_checkpointer_cm(config)
-    _checkpointer = _checkpointer_ctx.__enter__()
+            logger.info("Checkpointer: using InMemorySaver (in-process, not persistent)")
+            _checkpointer = InMemorySaver()
+            return _checkpointer
+
+        checkpointer_ctx = _sync_checkpointer_cm(config)
+        checkpointer = checkpointer_ctx.__enter__()
+        _checkpointer_ctx = checkpointer_ctx
+        _checkpointer = checkpointer
 
     return _checkpointer
 
@@ -155,13 +152,14 @@ def reset_checkpointer() -> None:
     Useful in tests or after a configuration change.
     """
     global _checkpointer, _checkpointer_ctx
-    if _checkpointer_ctx is not None:
-        try:
-            _checkpointer_ctx.__exit__(None, None, None)
-        except Exception:
-            logger.warning("Error during checkpointer cleanup", exc_info=True)
-        _checkpointer_ctx = None
-    _checkpointer = None
+    with _checkpointer_lock:
+        if _checkpointer_ctx is not None:
+            try:
+                _checkpointer_ctx.__exit__(None, None, None)
+            except Exception:
+                logger.warning("Error during checkpointer cleanup", exc_info=True)
+            _checkpointer_ctx = None
+        _checkpointer = None
 
 
 # ---------------------------------------------------------------------------

backend/packages/harness/deerflow/runtime/journal.py

@@ -164,7 +164,18 @@ class RunJournal(BaseCallbackHandler):
                 metadata={"caller": caller, **(metadata or {})},
             )
 
-    def on_chain_end(self, outputs: Any, *, run_id: UUID, **kwargs: Any) -> None:
+    def on_chain_end(
+        self,
+        outputs: Any,
+        *,
+        run_id: UUID,
+        parent_run_id: UUID | None = None,
+        **kwargs: Any,
+    ) -> None:
+        # Nested chain ends fire for internal graph nodes; only the root chain
+        # represents the user-visible run lifecycle.
+        if parent_run_id is not None:
+            return
         self._put(event_type="run.end", category="outputs", content=outputs, metadata={"status": "success"})
         self._flush_sync()
 

backend/packages/harness/deerflow/runtime/store/provider.py

@@ -22,11 +22,13 @@ from __future__ import annotations
 
 import contextlib
 import logging
+import threading
 from collections.abc import Iterator
 
 from langgraph.store.base import BaseStore
 
 from deerflow.config.app_config import get_app_config
+from deerflow.config.checkpointer_config import ensure_config_loaded
 from deerflow.runtime.store._sqlite_utils import ensure_sqlite_parent_dir, resolve_sqlite_conn_str
 
 logger = logging.getLogger(__name__)
@@ -100,6 +102,7 @@ def _sync_store_cm(config) -> Iterator[BaseStore]:
 
 _store: BaseStore | None = None
 _store_ctx = None  # open context manager keeping the connection alive
+_store_lock = threading.Lock()
 
 
 def get_store() -> BaseStore:
@@ -117,29 +120,29 @@ def get_store() -> BaseStore:
     if _store is not None:
         return _store
 
-    # Lazily load app config, mirroring the checkpointer singleton pattern so
-    # that tests that set the global checkpointer config explicitly remain isolated.
-    from deerflow.config.app_config import _app_config
-    from deerflow.config.checkpointer_config import get_checkpointer_config
+    # Config loading can reset both persistence singletons. Keep it outside
+    # this provider lock to avoid cross-provider lock-order inversion.
+    ensure_config_loaded()
 
-    config = get_checkpointer_config()
+    with _store_lock:
+        if _store is not None:
+            return _store
+
+        from deerflow.config.checkpointer_config import get_checkpointer_config
 
-    if config is None and _app_config is None:
-        try:
-            get_app_config()
-        except FileNotFoundError:
-            pass
         config = get_checkpointer_config()
 
-    if config is None:
-        from langgraph.store.memory import InMemoryStore
+        if config is None:
+            from langgraph.store.memory import InMemoryStore
 
-        logger.warning("No 'checkpointer' section in config.yaml — using InMemoryStore for the store. Thread list will be lost on server restart. Configure a sqlite or postgres backend for persistence.")
-        _store = InMemoryStore()
-        return _store
+            logger.warning("No 'checkpointer' section in config.yaml — using InMemoryStore for the store. Thread list will be lost on server restart. Configure a sqlite or postgres backend for persistence.")
+            _store = InMemoryStore()
+            return _store
 
-    _store_ctx = _sync_store_cm(config)
-    _store = _store_ctx.__enter__()
+        store_ctx = _sync_store_cm(config)
+        store = store_ctx.__enter__()
+        _store_ctx = store_ctx
+        _store = store
     return _store
 
 
@@ -150,13 +153,14 @@ def reset_store() -> None:
     Useful in tests or after a configuration change.
     """
     global _store, _store_ctx
-    if _store_ctx is not None:
-        try:
-            _store_ctx.__exit__(None, None, None)
-        except Exception:
-            logger.warning("Error during store cleanup", exc_info=True)
-        _store_ctx = None
-    _store = None
+    with _store_lock:
+        if _store_ctx is not None:
+            try:
+                _store_ctx.__exit__(None, None, None)
+            except Exception:
+                logger.warning("Error during store cleanup", exc_info=True)
+            _store_ctx = None
+        _store = None
 
 
 # ---------------------------------------------------------------------------

backend/packages/harness/deerflow/sandbox/local/local_sandbox_provider.py

@@ -147,7 +147,17 @@ class LocalSandboxProvider(SandboxProvider):
                             mount.container_path,
                         )
                         continue
-                    # Ensure the host path exists before adding mapping
+                    # Ensure the host path exists before adding mapping.
+                    #
+                    # ``host_path`` is resolved against the filesystem of the
+                    # process running this provider — for ``make dev`` that is
+                    # the host machine, but for ``make up`` it is the
+                    # ``deer-flow-gateway`` container, so any host path that
+                    # isn't bind-mounted into the gateway image will be missing
+                    # here. Skipping silently makes this a high-cost-to-debug
+                    # silent failure (sandbox skill / tool reads an empty dir
+                    # instead of the configured mount), so escalate to ERROR
+                    # and include actionable guidance. See #3244.
                     if host_path.exists():
                         mappings.append(
                             PathMapping(
@@ -157,10 +167,16 @@ class LocalSandboxProvider(SandboxProvider):
                             )
                         )
                     else:
-                        logger.warning(
-                            "Mount host_path does not exist, skipping: %s -> %s",
+                        logger.error(
+                            "sandbox.mounts entry %s -> %s ignored: host_path %s does not exist from the "
+                            "perspective of the gateway process. In Docker deployments (make up / docker-compose), "
+                            "this path must also be bind-mounted into the gateway container — add a matching "
+                            "volume entry under services.gateway.volumes in docker/docker-compose.yaml (and use "
+                            "the in-container path here), or run in local mode (make dev) where the gateway sees "
+                            "the host filesystem directly.",
                             mount.host_path,
                             mount.container_path,
+                            mount.host_path,
                         )
         except Exception as e:
             # Log but don't fail if config loading fails

backend/packages/harness/deerflow/skills/slash.py

@@ -0,0 +1,65 @@
+from __future__ import annotations
+
+import re
+from dataclasses import dataclass
+
+from deerflow.skills.types import Skill
+
+RESERVED_SLASH_SKILL_NAMES = frozenset({"bootstrap", "help", "memory", "models", "new", "status"})
+_SLASH_SKILL_RE = re.compile(r"^/([a-z0-9]+(?:-[a-z0-9]+)*)(?:\s+|$)")
+
+
+@dataclass(frozen=True, slots=True)
+class SlashSkillReference:
+    """Parsed slash-skill command with the skill name and remaining task text."""
+
+    name: str
+    remaining_text: str
+
+
+@dataclass(frozen=True, slots=True)
+class ResolvedSlashSkill:
+    """Slash-skill activation resolved against enabled runtime-visible skills."""
+
+    skill: Skill
+    remaining_text: str
+    container_file_path: str
+
+
+def parse_slash_skill_reference(text: str) -> SlashSkillReference | None:
+    """Parse strict `/skill-name task` syntax, ignoring reserved control commands."""
+    match = _SLASH_SKILL_RE.match(text)
+    if not match:
+        return None
+    name = match.group(1)
+    if name in RESERVED_SLASH_SKILL_NAMES:
+        return None
+    return SlashSkillReference(
+        name=name,
+        remaining_text=text[match.end() :].lstrip(),
+    )
+
+
+def resolve_slash_skill(
+    text: str,
+    skills: list[Skill],
+    *,
+    available_skills: set[str] | None = None,
+    container_base_path: str = "/mnt/skills",
+) -> ResolvedSlashSkill | None:
+    """Resolve text into an enabled, whitelisted skill activation if possible."""
+    reference = parse_slash_skill_reference(text)
+    if reference is None:
+        return None
+    if available_skills is not None and reference.name not in available_skills:
+        return None
+
+    skill = next((candidate for candidate in skills if candidate.name == reference.name and candidate.enabled), None)
+    if skill is None:
+        return None
+
+    return ResolvedSlashSkill(
+        skill=skill,
+        remaining_text=reference.remaining_text,
+        container_file_path=skill.get_container_file_path(container_base_path),
+    )

backend/packages/harness/deerflow/subagents/executor.py

@@ -12,7 +12,7 @@ from contextvars import Context, copy_context
 from dataclasses import dataclass, field
 from datetime import datetime
 from enum import Enum
-from typing import Any
+from typing import TYPE_CHECKING, Any
 
 from langchain.agents import create_agent
 from langchain.tools import BaseTool
@@ -28,6 +28,13 @@ from deerflow.skills.types import Skill
 from deerflow.subagents.config import SubagentConfig, resolve_subagent_model_name
 from deerflow.subagents.token_collector import SubagentTokenCollector
 
+if TYPE_CHECKING:
+    # Imported lazily at runtime inside _build_initial_state: importing
+    # tool_search eagerly would run tools/builtins/__init__ -> task_tool ->
+    # `from deerflow.subagents import SubagentExecutor`, which re-enters this
+    # still-initializing package. Type-only here keeps the annotation precise.
+    from deerflow.tools.builtins.tool_search import DeferredToolSetup
+
 logger = logging.getLogger(__name__)
 
 
@@ -319,8 +326,13 @@ class SubagentExecutor:
 
         logger.info(f"[trace={self.trace_id}] SubagentExecutor initialized: {config.name} with {len(self.tools)} tools")
 
-    def _create_agent(self, tools: list[BaseTool] | None = None):
-        """Create the agent instance."""
+    def _create_agent(self, tools: list[BaseTool] | None = None, *, deferred_setup: "DeferredToolSetup | None" = None):
+        """Create the agent instance.
+
+        ``deferred_setup`` (assembled in ``_build_initial_state``) carries the
+        deferred MCP tool names + catalog hash so the subagent gets the same
+        DeferredToolFilterMiddleware the lead agent has. ``None`` is a no-op.
+        """
         app_config = self.app_config or get_app_config()
         if self.model_name is None:
             self.model_name = resolve_subagent_model_name(self.config, self.parent_model, app_config=app_config)
@@ -329,7 +341,7 @@ class SubagentExecutor:
         from deerflow.agents.middlewares.tool_error_handling_middleware import build_subagent_runtime_middlewares
 
         # Reuse shared middleware composition with lead agent.
-        middlewares = build_subagent_runtime_middlewares(app_config=app_config, model_name=self.model_name, lazy_init=True)
+        middlewares = build_subagent_runtime_middlewares(app_config=app_config, model_name=self.model_name, lazy_init=True, deferred_setup=deferred_setup)
 
         # system_prompt is included in initial state messages (see _build_initial_state)
         # to avoid multiple SystemMessages which some LLM APIs don't support.
@@ -403,19 +415,35 @@ class SubagentExecutor:
 
         return messages
 
-    async def _build_initial_state(self, task: str) -> tuple[dict[str, Any], list[BaseTool]]:
+    async def _build_initial_state(self, task: str) -> tuple[dict[str, Any], list[BaseTool], "DeferredToolSetup"]:
         """Build the initial state for agent execution.
 
         Args:
             task: The task description.
 
         Returns:
-            Initial state dictionary and tools filtered by loaded skill metadata.
+            ``(state, final_tools, deferred_setup)``. ``final_tools`` is the
+            policy-filtered tool list with the ``tool_search`` tool appended when
+            deferral applies; ``deferred_setup`` is consumed by ``_create_agent``
+            so the agent build and the injected ``<available-deferred-tools>``
+            section share one catalog/hash.
         """
+        # Lazy import: see the TYPE_CHECKING note at the top of this module -
+        # importing tool_search runs tools/builtins/__init__, which would
+        # re-enter this package during its own initialization.
+        from deerflow.tools.builtins.tool_search import assemble_deferred_tools, get_deferred_tools_prompt_section
 
         # Load skills as conversation items (Codex pattern)
         skills = await self._load_skills()
         filtered_tools = self._apply_skill_allowed_tools(skills)
+        # Assemble deferred tool_search AFTER policy filtering (fail-closed),
+        # mirroring the lead path so subagents stop binding full MCP schemas.
+        # The generated tool_search helper is intentionally not subject to the
+        # subagent's name-level allow/deny (config.tools / disallowed_tools):
+        # its catalog is built from the already-filtered list, so it can never
+        # surface a tool the policy denied. This matches the lead agent.
+        enabled = (self.app_config or get_app_config()).tool_search.enabled
+        final_tools, deferred_setup = assemble_deferred_tools(filtered_tools, enabled=enabled)
         skill_messages = await self._load_skill_messages(skills)
 
         # Combine system_prompt and skills into a single SystemMessage.
@@ -426,6 +454,11 @@ class SubagentExecutor:
             system_parts.append(self.config.system_prompt)
         for skill_msg in skill_messages:
             system_parts.append(skill_msg.content)
+        # Name the deferred MCP tools in the prompt; their schemas stay withheld
+        # until tool_search promotes them. Empty set -> "" -> appends nothing.
+        deferred_section = get_deferred_tools_prompt_section(deferred_names=deferred_setup.deferred_names)
+        if deferred_section:
+            system_parts.append(deferred_section)
 
         messages: list[Any] = []
         if system_parts:
@@ -444,7 +477,7 @@ class SubagentExecutor:
         if self.thread_data is not None:
             state["thread_data"] = self.thread_data
 
-        return state, filtered_tools
+        return state, final_tools, deferred_setup
 
     async def _aexecute(self, task: str, result_holder: SubagentResult | None = None) -> SubagentResult:
         """Execute a task asynchronously.
@@ -475,8 +508,8 @@ class SubagentExecutor:
 
         collector: SubagentTokenCollector | None = None
         try:
-            state, filtered_tools = await self._build_initial_state(task)
-            agent = self._create_agent(filtered_tools)
+            state, final_tools, deferred_setup = await self._build_initial_state(task)
+            agent = self._create_agent(final_tools, deferred_setup=deferred_setup)
 
             # Token collector for subagent LLM calls
             collector_caller = f"subagent:{self.config.name}"

backend/packages/harness/deerflow/subagents/status_contract.py

@@ -0,0 +1,102 @@
+"""Backend↔frontend contract for the structured subagent status.
+
+Bytedance/deer-flow issue #3146: the frontend used to derive the
+subtask card state by string-matching the leading text of the
+``task`` tool's result. That contract was fragile — any rewording on
+the backend silently broke the card lifecycle, and the issue history
+of #3107 BUG-007 / #3131 review showed it repeatedly.
+
+This module replaces the text-shaped contract with a small structured
+one carried inside ``ToolMessage.additional_kwargs``:
+
+- ``subagent_status``: one of ``SUBAGENT_STATUS_VALUES``.
+- ``subagent_error`` (optional): the human-readable error blob the
+  backend recorded.
+
+The mapping from "task tool result text" to status is the one piece
+the backend stamper (``ToolErrorHandlingMiddleware``) and the
+frontend fallback parser must agree on. The shared fixture at
+``contracts/subagent_status_contract.json`` is the single source of
+truth — both sides' tests load it and assert behaviour.
+"""
+
+from __future__ import annotations
+
+from typing import Literal
+
+SUBAGENT_STATUS_KEY = "subagent_status"
+SUBAGENT_ERROR_KEY = "subagent_error"
+
+SubagentStatusValue = Literal[
+    "completed",
+    "failed",
+    "cancelled",
+    "timed_out",
+    "polling_timed_out",
+]
+
+#: Enumeration of every value ``subagent_status`` may take. Mirrors the
+#: ``valid_status_values`` array in the shared fixture; the contract test
+#: pins them against each other.
+SUBAGENT_STATUS_VALUES: tuple[SubagentStatusValue, ...] = (
+    "completed",
+    "failed",
+    "cancelled",
+    "timed_out",
+    "polling_timed_out",
+)
+
+# Prefix table — ordered most-specific-first because some prefixes are
+# substrings of others ("Task timed out" vs "Task polling timed out", "Task
+# failed" vs "Task failed. Error: ..."). The "Task " prefixes come from
+# ``task_tool.py``'s 5 normal-return strings; the bare ``Error:`` prefix
+# catches both the 3 ``Error:`` pre-execution returns and the wrapper
+# produced by ``ToolErrorHandlingMiddleware`` for any task tool exception.
+_PREFIX_TO_STATUS: tuple[tuple[str, SubagentStatusValue], ...] = (
+    ("Task Succeeded. Result:", "completed"),
+    ("Task polling timed out", "polling_timed_out"),
+    ("Task timed out", "timed_out"),
+    ("Task cancelled by user", "cancelled"),
+    ("Task failed.", "failed"),
+    ("Error", "failed"),
+)
+
+
+def extract_subagent_status(content: str) -> SubagentStatusValue | None:
+    """Infer the structured status for a ``task`` tool result string.
+
+    Returns ``None`` when the content does not match any known terminal
+    prefix. Non-terminal streaming chunks fall into this branch by
+    design — the middleware then leaves ``subagent_status`` unset so
+    the frontend keeps the card on its in-progress placeholder until
+    the real terminal frame arrives.
+    """
+    trimmed = content.strip()
+    for prefix, status in _PREFIX_TO_STATUS:
+        if trimmed.startswith(prefix):
+            return status
+    return None
+
+
+def make_subagent_additional_kwargs(
+    status: SubagentStatusValue,
+    *,
+    error: str | None = None,
+) -> dict[str, str]:
+    """Build the ``additional_kwargs`` payload the middleware stamps.
+
+    Drops the error field when blank so the JSON wire format never carries
+    a misleading empty ``subagent_error: ""``.
+
+    Raises:
+        ValueError: when ``status`` is not in :data:`SUBAGENT_STATUS_VALUES`.
+            We do not accept arbitrary strings: a typo would silently leak
+            through to the frontend and degrade to the legacy prefix
+            fallback rather than failing loudly.
+    """
+    if status not in SUBAGENT_STATUS_VALUES:
+        raise ValueError(f"invalid subagent status {status!r}; expected one of {SUBAGENT_STATUS_VALUES}")
+    payload: dict[str, str] = {SUBAGENT_STATUS_KEY: status}
+    if error and error.strip():
+        payload[SUBAGENT_ERROR_KEY] = error.strip()
+    return payload

backend/packages/harness/deerflow/tools/builtins/tool_search.py

@@ -179,3 +179,43 @@ def build_deferred_tool_setup(filtered_tools: list[BaseTool], *, enabled: bool)
         return DeferredToolSetup(None, frozenset(), None)
     catalog = DeferredToolCatalog(tuple(deferred))
     return DeferredToolSetup(build_tool_search_tool(catalog), catalog.names, catalog.hash)
+
+
+def assemble_deferred_tools(filtered_tools: list[BaseTool], *, enabled: bool) -> tuple[list[BaseTool], DeferredToolSetup]:
+    """Build the final tool list + deferred setup from a POLICY-FILTERED list.
+
+    Call AFTER tool-policy filtering so the deferred catalog never exposes a tool
+    the agent is not allowed to use. Fail-closed: if tool_search is enabled and
+    MCP tools survived filtering but no deferred set was recovered, raise rather
+    than silently binding their full schemas to the model.
+
+    Shared by every agent-build path (lead, embedded client, subagent) so they
+    all get the same fail-closed guarantee from one place.
+    """
+    deferred_setup = build_deferred_tool_setup(filtered_tools, enabled=enabled)
+    if enabled and not deferred_setup.deferred_names and any(is_mcp_tool(t) for t in filtered_tools):
+        raise RuntimeError("tool_search enabled and MCP tools survived policy filtering, but no deferred set was recovered - refusing to bind MCP schemas (fail-closed).")
+    final_tools = list(filtered_tools)
+    if deferred_setup.tool_search_tool:
+        final_tools.append(deferred_setup.tool_search_tool)
+    return final_tools, deferred_setup
+
+
+# Prompt rendering
+
+
+def get_deferred_tools_prompt_section(*, deferred_names: frozenset[str] = frozenset()) -> str:
+    """Generate <available-deferred-tools> from an explicit deferred-name set.
+
+    Lists only names so the agent knows what exists and can use tool_search to
+    load them. Returns empty string when there are no deferred tools. The set is
+    computed at agent build time (after tool-policy filtering) and passed in.
+
+    Lives here, next to the assembly that produces ``deferred_names``, so every
+    agent-build path (lead, embedded client, subagent) renders the section the
+    same way without coupling back to ``lead_agent.prompt``.
+    """
+    if not deferred_names:
+        return ""
+    names = "\n".join(sorted(deferred_names))
+    return f"<available-deferred-tools>\n{names}\n</available-deferred-tools>"

backend/packages/harness/deerflow/utils/messages.py

@@ -0,0 +1,31 @@
+from __future__ import annotations
+
+from collections.abc import Mapping
+from typing import Any
+
+ORIGINAL_USER_CONTENT_KEY = "original_user_content"
+
+
+def message_content_to_text(content: Any) -> str:
+    """Extract text from LangChain message content shapes."""
+    if isinstance(content, str):
+        return content
+    if isinstance(content, list):
+        parts: list[str] = []
+        for item in content:
+            if isinstance(item, str):
+                parts.append(item)
+            elif isinstance(item, dict):
+                text = item.get("text")
+                if isinstance(text, str):
+                    parts.append(text)
+        return "\n".join(part for part in parts if part)
+    return str(content)
+
+
+def get_original_user_content_text(content: Any, additional_kwargs: Mapping[str, Any] | None) -> str:
+    """Return pre-middleware user text when available, otherwise content text."""
+    original_content = (additional_kwargs or {}).get(ORIGINAL_USER_CONTENT_KEY)
+    if isinstance(original_content, str):
+        return original_content
+    return message_content_to_text(content)

backend/packages/harness/pyproject.toml

@@ -36,6 +36,7 @@ dependencies = [
     "sqlalchemy[asyncio]>=2.0,<3.0",
     "aiosqlite>=0.19",
     "alembic>=1.13",
+    "cryptography>=43.0.0",
 ]
 
 [project.optional-dependencies]

backend/scripts/build_fixture_from_jsonl.py

@@ -0,0 +1,45 @@
+"""Turn a record-through-browser JSONL capture into a replay fixture.
+
+The recording gateway (``record_gateway.py``) appends ``{input_hash, output}``
+lines as the frontend drives a real run; the record spec writes a ``.meta.json``
+sidecar with ``{scenario, mode, prompt}``. This stitches them into the fixture
+the replay provider + tests consume.
+"""
+
+from __future__ import annotations
+
+import argparse
+import json
+from pathlib import Path
+
+
+def main() -> int:
+    parser = argparse.ArgumentParser()
+    parser.add_argument("--jsonl", required=True)
+    parser.add_argument("--meta", required=True)
+    parser.add_argument("--out", required=True)
+    parser.add_argument("--model", default="gpt-5.5")
+    args = parser.parse_args()
+
+    turns = [json.loads(line) for line in Path(args.jsonl).read_text(encoding="utf-8").splitlines() if line.strip()]
+    meta = json.loads(Path(args.meta).read_text(encoding="utf-8"))
+    fixture = {
+        "scenario": meta["scenario"],
+        "mode": meta["mode"],
+        "model": args.model,
+        "prompt": meta["prompt"],
+        "context": meta.get("context", {}),
+        "turns": turns,
+    }
+    Path(args.out).write_text(json.dumps(fixture, ensure_ascii=False, indent=2), encoding="utf-8")
+    print(f"wrote {len(turns)} turn(s) -> {args.out}")
+    for index, turn in enumerate(turns):
+        data = turn["output"].get("data", {})
+        tool_calls = [tc.get("name") for tc in (data.get("tool_calls") or [])]
+        caller = turn.get("caller", "legacy")
+        print(f"  turn {index}: caller={caller} hash={turn['input_hash'][:12]} tool_calls={tool_calls} content={str(data.get('content'))[:50]!r}")
+    return 0
+
+
+if __name__ == "__main__":
+    raise SystemExit(main())

backend/scripts/record_gateway.py

@@ -0,0 +1,127 @@
+"""Recording gateway for *record-through-browser* (Plan A).
+
+Runs the gateway with a REAL model and a callback that appends every model
+call's ``(input_hash, output)`` to a JSONL file. Because the run is driven by
+the real frontend (Playwright), the captured inputs are EXACTLY what the
+frontend produces (date system-reminder, suggestions/title calls, ...), so the
+resulting fixture replays cleanly against the browser.
+
+Used by ``frontend/playwright.record.config.ts``. Env:
+  OPENAI_API_KEY / OPENAI_API_BASE  - the real upstream (never committed)
+  DEERFLOW_RECORD_OUT               - JSONL path to append captured turns to
+  RECORD_PORT (default 8012), RECORD_MODEL (default gpt-5.5)
+"""
+
+from __future__ import annotations
+
+import json
+import os
+import sys
+import tempfile
+from pathlib import Path
+
+_BACKEND = Path(__file__).resolve().parents[1]
+sys.path.insert(0, str(_BACKEND))
+sys.path.insert(0, str(_BACKEND / "tests"))
+
+
+def _install_capture(out_path: Path) -> None:
+    from langchain_core.callbacks import BaseCallbackHandler
+    from langchain_core.messages import messages_to_dict
+    from replay_provider import caller_identity, hash_messages, hash_replay_input
+
+    import deerflow.models.factory as factory_mod
+
+    class Capture(BaseCallbackHandler):
+        def __init__(self) -> None:
+            self.inputs: dict[str, tuple[list, str]] = {}
+
+        def on_chat_model_start(  # noqa: ANN001
+            self,
+            serialized,
+            messages,
+            *,
+            run_id=None,
+            tags=None,
+            name=None,
+            **kwargs,
+        ):
+            self.inputs[str(run_id)] = (
+                messages[0] if messages else [],
+                caller_identity(name=name, tags=tags),
+            )
+
+        def on_llm_end(self, response, *, run_id=None, **kwargs):  # noqa: ANN001
+            captured = self.inputs.pop(str(run_id), None)
+            if captured is None:
+                return
+            inp, caller = captured
+            for batch in response.generations:
+                for gen in batch:
+                    message = getattr(gen, "message", None)
+                    if message is None:
+                        continue
+                    record = {
+                        "caller": caller,
+                        "conversation_hash": hash_messages(inp),
+                        "input_hash": hash_replay_input(inp, caller=caller),
+                        "output": messages_to_dict([message])[0],
+                    }
+                    with open(out_path, "a", encoding="utf-8") as handle:
+                        handle.write(json.dumps(record, ensure_ascii=False) + "\n")
+                        handle.flush()
+
+    cb = Capture()
+    original = factory_mod.create_chat_model
+
+    def wrapped(*args, **kwargs):
+        model = original(*args, **kwargs)
+        model.callbacks = (model.callbacks or []) + [cb]
+        return model
+
+    factory_mod.create_chat_model = wrapped
+    for module in list(sys.modules.values()):
+        if getattr(module, "create_chat_model", None) is original:
+            module.create_chat_model = wrapped
+
+
+def main() -> int:
+    if not os.environ.get("OPENAI_API_KEY") or not os.environ.get("OPENAI_API_BASE"):
+        print("ERROR: set OPENAI_API_KEY and OPENAI_API_BASE (an OpenAI-compatible /v1 endpoint)", file=sys.stderr)
+        return 2
+
+    record_out = os.environ.get("DEERFLOW_RECORD_OUT")
+    if not record_out:
+        print("ERROR: set DEERFLOW_RECORD_OUT to the JSONL path to append captured turns to", file=sys.stderr)
+        return 2
+
+    port = int(os.environ.get("RECORD_PORT", "8012"))
+    model = os.environ.get("RECORD_MODEL", "gpt-5.5")
+    out = Path(record_out)
+    out.parent.mkdir(parents=True, exist_ok=True)
+    out.write_text("", encoding="utf-8")  # fresh capture per recording run
+
+    from _replay_fixture import build_config_yaml, prepare_hermetic_extras, real_model_block
+
+    home = Path(tempfile.mkdtemp(prefix="record-gw-"))
+    cfg = home / "config.yaml"
+    cfg.write_text(build_config_yaml(model_block=real_model_block(model), home=home), encoding="utf-8")
+    # Override (not setdefault): the recorder must be hermetic, so an outer
+    # DEER_FLOW_HOME can't leak in and shift prompt-affecting paths/skills.
+    os.environ["DEER_FLOW_HOME"] = str(home)
+    os.environ["DEER_FLOW_CONFIG_PATH"] = str(cfg)
+    os.environ["DEER_FLOW_EXTENSIONS_CONFIG_PATH"] = str(prepare_hermetic_extras(home))
+    os.environ.setdefault("AUTH_JWT_SECRET", "record-secret")
+    os.environ["PYTHONPATH"] = os.pathsep.join(p for p in (str(_BACKEND), str(_BACKEND / "tests"), os.environ.get("PYTHONPATH", "")) if p)
+
+    _install_capture(out)
+
+    import uvicorn
+
+    print(f"[record-gw] model={model} out={out} port={port}", flush=True)
+    uvicorn.run("app.gateway.app:app", host="127.0.0.1", port=port, log_level="warning")
+    return 0
+
+
+if __name__ == "__main__":
+    raise SystemExit(main())

backend/scripts/run_replay_gateway.py

@@ -0,0 +1,73 @@
+"""Start a hermetic *replay* gateway for the full-stack (Layer 2) e2e.
+
+Builds an ephemeral config that points the model at ``ReplayChatModel`` + a
+recorded fixture, then runs uvicorn — no API key, deterministic. Used as a
+Playwright ``webServer`` (see ``frontend/playwright.real-backend.config.ts``) and
+runnable standalone for debugging::
+
+    uv run python scripts/run_replay_gateway.py --port 8011
+
+``tests/`` is put on the path so the config ``use: replay_provider:ReplayChatModel``
+resolves; ``GATEWAY_CORS_ORIGINS`` is set so the frontend on :3000 can talk to it.
+"""
+
+from __future__ import annotations
+
+import argparse
+import os
+import sys
+import tempfile
+from pathlib import Path
+
+_BACKEND = Path(__file__).resolve().parents[1]
+sys.path.insert(0, str(_BACKEND))
+sys.path.insert(0, str(_BACKEND / "tests"))  # replay_provider + build_config_yaml live here
+
+
+def main() -> int:
+    parser = argparse.ArgumentParser()
+    parser.add_argument("--port", type=int, default=8011)
+    parser.add_argument("--fixture", default=str(_BACKEND / "tests" / "fixtures" / "replay" / "write_read_file.ultra.json"))
+    parser.add_argument("--cors", default="http://localhost:3000")
+    args = parser.parse_args()
+
+    from _replay_fixture import REPLAY_MODEL_BLOCK, build_config_yaml, prepare_hermetic_extras
+
+    home = Path(tempfile.mkdtemp(prefix="replay-gw-"))
+    cfg = home / "config.yaml"
+    cfg.write_text(build_config_yaml(model_block=REPLAY_MODEL_BLOCK, home=home), encoding="utf-8")
+
+    # Override (not setdefault): the replay gateway must be hermetic, so an outer
+    # DEER_FLOW_HOME can't leak in and shift prompt-affecting paths/skills.
+    os.environ["DEER_FLOW_HOME"] = str(home)
+    os.environ["DEER_FLOW_CONFIG_PATH"] = str(cfg)
+    os.environ["DEER_FLOW_EXTENSIONS_CONFIG_PATH"] = str(prepare_hermetic_extras(home))
+    os.environ["DEERFLOW_REPLAY_FIXTURE"] = args.fixture
+    os.environ.setdefault("AUTH_JWT_SECRET", "ci-replay-secret")
+    os.environ["GATEWAY_CORS_ORIGINS"] = args.cors
+    # Child / dynamic imports (resolve_class) search PYTHONPATH too.
+    os.environ["PYTHONPATH"] = os.pathsep.join(p for p in (str(_BACKEND), str(_BACKEND / "tests"), os.environ.get("PYTHONPATH", "")) if p)
+
+    import uvicorn
+
+    target: str | object = "app.gateway.app:app"
+    # Test-only: attach the run/message seeder used by the multi-run render-order
+    # e2e (#3352). Imported from tests/ and mounted here only — never in the
+    # production app. Pass the app object (not the import string) so the extra
+    # router is registered before uvicorn serves it.
+    if os.environ.get("DEERFLOW_ENABLE_TEST_SEED") == "1":
+        from seed_runs_router import router as seed_router
+
+        from app.gateway.app import app as gateway_app
+
+        gateway_app.include_router(seed_router)
+        target = gateway_app
+        print("[replay-gw] test-only seed router mounted at /api/test-only/seed-runs", flush=True)
+
+    print(f"[replay-gw] config={cfg} fixture={args.fixture} cors={args.cors} port={args.port}", flush=True)
+    uvicorn.run(target, host="127.0.0.1", port=args.port, log_level="warning")
+    return 0
+
+
+if __name__ == "__main__":
+    raise SystemExit(main())

backend/sitecustomize.py

@@ -0,0 +1,26 @@
+"""Process-wide Python startup customizations for backend entrypoints.
+
+When ``backend/`` is on ``sys.path``, Python imports this module during
+interpreter startup. Keep changes here suitable for all gateway, script,
+migration, and test entrypoints that run in that environment.
+"""
+
+from __future__ import annotations
+
+import asyncio
+import sys
+
+
+def _configure_windows_event_loop_policy() -> None:
+    if sys.platform != "win32":
+        return
+
+    selector_policy = getattr(asyncio, "WindowsSelectorEventLoopPolicy", None)
+    if selector_policy is None:
+        return
+
+    if not isinstance(asyncio.get_event_loop_policy(), selector_policy):
+        asyncio.set_event_loop_policy(selector_policy())
+
+
+_configure_windows_event_loop_policy()

backend/tests/_replay_fixture.py

@@ -0,0 +1,164 @@
+"""Shared config + gateway-drive helpers for the record/replay e2e.
+
+Record (``scripts/record_gateway.py`` + ``scripts/build_fixture_from_jsonl.py``)
+and replay (``tests/test_replay_golden.py``)
+MUST drive the gateway through an identical, prompt-affecting config — otherwise
+the system prompt differs and the recorded input hashes never match on replay.
+Centralising the config builder + drive loop here makes that identity hold by
+construction; only the ``models[].use`` block differs (real model vs
+``ReplayChatModel``).
+"""
+
+from __future__ import annotations
+
+import json
+import uuid
+from pathlib import Path
+
+# mode -> (thinking_enabled, is_plan_mode, subagent_enabled). Mirrors the
+# frontend mapping in core/threads/hooks.ts.
+MODE_CONTEXT: dict[str, tuple[bool, bool, bool]] = {
+    "flash": (False, False, False),
+    "thinking": (True, False, False),
+    "pro": (True, True, False),
+    # thinking_enabled mirrors the frontend `context.mode !== "flash"` (hooks.ts),
+    # so ultra is thinking-enabled too.
+    "ultra": (True, True, True),
+}
+
+# The replay model block: same model NAME as recording (so nothing in the prompt
+# shifts), only ``use`` swapped to the deterministic replay provider.
+REPLAY_MODEL_BLOCK = """\
+  - name: scenario-model
+    display_name: Scenario Model
+    use: replay_provider:ReplayChatModel
+    model: replay
+    supports_thinking: true"""
+
+
+def real_model_block(model: str) -> str:
+    return f"""\
+  - name: scenario-model
+    display_name: Scenario Model
+    use: langchain_openai:ChatOpenAI
+    model: {model}
+    api_key: $OPENAI_API_KEY
+    base_url: $OPENAI_API_BASE"""
+
+
+def build_config_yaml(*, model_block: str, home: Path) -> str:
+    """Full gateway config. Only ``model_block`` varies between record/replay.
+
+    Everything that shapes the system prompt is pinned so record, replay, and CI
+    produce byte-identical prompts regardless of the machine:
+    - sandbox / tool_groups / tools — fixed here
+    - skills — pointed at an empty ``<home>/skills`` so filesystem skills (incl.
+      gitignored custom skills present only on a dev box) never leak into the
+      prompt. Pair with an empty ``extensions_config.json`` (no MCP) via
+      :func:`prepare_hermetic_extras`.
+    - memory / summarization — disabled (background, non-deterministic timing)
+    """
+    return f"""\
+log_level: warning
+models:
+{model_block}
+sandbox:
+  use: deerflow.sandbox.local:LocalSandboxProvider
+skills:
+  path: {home / "skills"}
+  container_path: /mnt/skills
+tool_groups:
+  - name: file:read
+  - name: file:write
+tools:
+  - name: ls
+    group: file:read
+    use: deerflow.sandbox.tools:ls_tool
+  - name: read_file
+    group: file:read
+    use: deerflow.sandbox.tools:read_file_tool
+  - name: write_file
+    group: file:write
+    use: deerflow.sandbox.tools:write_file_tool
+# Memory + summarization make background / debounced model calls whose timing is
+# non-deterministic; disable them so record and replay see the same model-call
+# set. (Title stays — it is an in-graph, deterministic call we record.)
+memory:
+  enabled: false
+  injection_enabled: false
+summarization:
+  enabled: false
+agents_api:
+  enabled: true
+database:
+  backend: sqlite
+  sqlite_dir: {home / "db"}
+"""
+
+
+def prepare_hermetic_extras(home: Path) -> Path:
+    """Create the empty skills tree + an empty extensions_config.json so the
+    system prompt has no environment-dependent skills/MCP content.
+
+    Returns the extensions-config path; the caller must point
+    ``DEER_FLOW_EXTENSIONS_CONFIG_PATH`` at it. Call before starting the gateway.
+    """
+    (home / "skills" / "public").mkdir(parents=True, exist_ok=True)
+    (home / "skills" / "custom").mkdir(parents=True, exist_ok=True)
+    extensions = home / "extensions_config.json"
+    extensions.write_text(json.dumps({"mcpServers": {}, "skills": {}}), encoding="utf-8")
+    return extensions
+
+
+def sse_event_shapes(resp) -> list[dict]:
+    """Reduce an SSE stream to (event name, sorted top-level data keys).
+
+    Snapshots the *shape* of the stream, not volatile values, so the golden is
+    stable across runs while still catching event-sequence / payload-shape drift.
+    """
+    events: list[dict] = []
+    current: str | None = None
+    for line in resp.iter_lines():
+        if line.startswith("event:"):
+            current = line[len("event:") :].strip()
+        elif line.startswith("data:"):
+            raw = line[len("data:") :].strip()
+            try:
+                data = json.loads(raw) if raw else {}
+            except json.JSONDecodeError:
+                data = {"_raw": raw[:200]}
+            events.append({"event": current, "keys": sorted(data.keys()) if isinstance(data, dict) else None})
+    return events
+
+
+def drive_gateway(app, *, prompt: str, context: dict) -> list[dict]:
+    """Register -> create thread -> POST /runs/stream; return SSE event shapes.
+
+    This is the exact wire path the React frontend uses (LangGraph SDK), driven
+    in-process via Starlette's TestClient with the real auth flow.
+    """
+    from starlette.testclient import TestClient
+
+    with TestClient(app) as client:
+        reg = client.post(
+            "/api/v1/auth/register",
+            json={"email": f"e2e-{uuid.uuid4().hex[:8]}@example.com", "password": "very-strong-password-123"},
+        )
+        assert reg.status_code == 201, reg.text
+        csrf = client.cookies.get("csrf_token")
+        assert csrf, "register must set csrf_token cookie"
+
+        thread_id = str(uuid.uuid4())
+        created = client.post("/api/threads", json={"thread_id": thread_id, "metadata": {}}, headers={"X-CSRF-Token": csrf})
+        assert created.status_code == 200, created.text
+
+        body = {
+            "assistant_id": "lead_agent",
+            "input": {"messages": [{"role": "user", "content": prompt}]},
+            "config": {"recursion_limit": 50},
+            "context": context,
+            "stream_mode": ["values"],
+        }
+        with client.stream("POST", f"/api/threads/{thread_id}/runs/stream", json=body, headers={"X-CSRF-Token": csrf}) as resp:
+            assert resp.status_code == 200, resp.read().decode()
+            return sse_event_shapes(resp)

backend/tests/blocking_io/test_agents_router.py

@@ -0,0 +1,64 @@
+"""Regression anchors: the custom-agent router must not block the event loop.
+
+``app.gateway.routers.agents.create_agent_endpoint`` and ``delete_agent`` are
+async route handlers that resolve the agent directory (``Paths.base_dir`` calls
+``Path.resolve``), probe it (``Path.exists``), and create/remove it (``mkdir``,
+config/SOUL writes, ``shutil.rmtree``) — all blocking IO. Both offload that work
+via ``asyncio.to_thread``; if any of it regresses back onto the event loop, the
+strict Blockbuster gate raises ``BlockingError`` and these tests fail.
+
+Imports live at module scope so the one-time FastAPI app construction (which
+reads files while building OpenAPI schemas) happens at collection time, not on
+the event loop under test. Test-side path resolution is itself offloaded with
+``asyncio.to_thread`` (matching ``test_uploads_middleware``) so only the
+handlers' own filesystem access is exercised on the loop.
+"""
+
+from __future__ import annotations
+
+import asyncio
+from pathlib import Path
+
+import pytest
+
+from app.gateway.routers.agents import AgentCreateRequest, create_agent_endpoint, delete_agent
+from deerflow.config.agents_api_config import load_agents_api_config_from_dict
+from deerflow.config.paths import get_paths
+from deerflow.runtime.user_context import get_effective_user_id
+
+pytestmark = pytest.mark.asyncio
+
+
+async def test_create_agent_does_not_block_event_loop(tmp_path: Path, monkeypatch) -> None:
+    monkeypatch.setenv("DEER_FLOW_HOME", str(tmp_path))
+    monkeypatch.setattr("deerflow.config.paths._paths", None)
+    load_agents_api_config_from_dict({"enabled": True})
+    try:
+        response = await create_agent_endpoint(AgentCreateRequest(name="loop-make-agent", soul="You are a test agent."))
+        assert response is not None
+
+        user_id = get_effective_user_id()
+        # test-side check (resolution offloaded; not exercised on the loop)
+        agent_dir = await asyncio.to_thread(get_paths().user_agent_dir, user_id, "loop-make-agent")
+        assert await asyncio.to_thread((agent_dir / "config.yaml").exists)
+    finally:
+        load_agents_api_config_from_dict({})
+
+
+async def test_delete_agent_does_not_block_event_loop(tmp_path: Path, monkeypatch) -> None:
+    monkeypatch.setenv("DEER_FLOW_HOME", str(tmp_path))
+    monkeypatch.setattr("deerflow.config.paths._paths", None)
+    load_agents_api_config_from_dict({"enabled": True})
+    try:
+        user_id = get_effective_user_id()
+        user_id = get_effective_user_id()
+        # test-side seeding (resolution offloaded; not exercised on the loop)
+        agent_dir = await asyncio.to_thread(get_paths().user_agent_dir, user_id, "loop-test-agent")
+        await asyncio.to_thread(agent_dir.mkdir, parents=True, exist_ok=True)
+        await asyncio.to_thread((agent_dir / "config.yaml").write_text, "name: loop-test-agent\n", encoding="utf-8")
+
+        await delete_agent("loop-test-agent")
+
+        assert not await asyncio.to_thread(agent_dir.exists)
+    finally:
+        load_agents_api_config_from_dict({})

backend/tests/blocking_io/test_dynamic_context_middleware.py

@@ -0,0 +1,124 @@
+"""Regression anchor: DynamicContextMiddleware must not block the event loop.
+
+``_inject`` performs synchronous file I/O (memory JSON loading) and
+potentially blocking network calls (tiktoken encoding download on first
+use — see issue #3402).  ``abefore_agent`` offloads the call via
+``asyncio.to_thread`` so the event loop stays responsive.
+
+This anchor drives the real ``create_agent`` graph via ``ainvoke`` under
+the strict Blockbuster gate.  If the offload regresses and the blocking
+I/O runs on the event loop, Blockbuster raises ``BlockingError`` and
+this test fails.
+"""
+
+from __future__ import annotations
+
+import asyncio
+from types import SimpleNamespace
+from unittest import mock
+
+import pytest
+from langchain.agents import create_agent
+from langchain_core.language_models.fake_chat_models import FakeMessagesListChatModel
+from langchain_core.messages import AIMessage, HumanMessage
+
+from deerflow.agents.middlewares.dynamic_context_middleware import DynamicContextMiddleware
+
+pytestmark = pytest.mark.asyncio
+
+
+class _FakeModel(FakeMessagesListChatModel):
+    """FakeMessagesListChatModel with a no-op ``bind_tools`` for create_agent."""
+
+    def bind_tools(self, tools, **kwargs):  # type: ignore[override]
+        return self
+
+
+async def test_abefore_agent_does_not_block_event_loop() -> None:
+    """``abefore_agent`` must offload _inject() to a thread pool."""
+    mw = DynamicContextMiddleware()
+
+    # Mock _build_full_reminder to simulate a slow synchronous operation
+    # (file I/O + tiktoken download).  The mock sleeps briefly to make any
+    # event-loop blocking visible to the Blockbuster gate.
+    original_build = mw._build_full_reminder
+
+    def slow_build_reminder():
+        import time
+
+        time.sleep(0.05)  # 50ms sync sleep — blocks the thread it runs on
+        return original_build()
+
+    with (
+        mock.patch.object(mw, "_build_full_reminder", slow_build_reminder),
+        mock.patch("deerflow.agents.lead_agent.prompt._get_memory_context", return_value=""),
+    ):
+        agent = await asyncio.to_thread(
+            lambda: create_agent(
+                model=_FakeModel(responses=[AIMessage(content="ok")]),
+                tools=[],
+                middleware=[mw],
+            )
+        )
+
+        result = await agent.ainvoke(
+            {"messages": [HumanMessage(content="hi")]},
+            {"configurable": {"thread_id": "test-thread"}},
+        )
+
+    assert result["messages"]
+
+
+async def test_abefore_agent_returns_same_result_as_before_agent() -> None:
+    """``abefore_agent`` (async, offloaded) must produce the same result as
+    ``before_agent`` (sync, for backward compatibility)."""
+    mw = DynamicContextMiddleware()
+
+    state = {"messages": [HumanMessage(content="Hello", id="msg-1")]}
+    runtime = SimpleNamespace(context={})
+
+    with (
+        mock.patch("deerflow.agents.lead_agent.prompt._get_memory_context", return_value=""),
+        mock.patch("deerflow.agents.middlewares.dynamic_context_middleware.datetime") as mock_dt,
+    ):
+        mock_dt.now.return_value.strftime.return_value = "2026-06-05, Friday"
+
+        # Sync path
+        sync_result = mw.before_agent(state, runtime)
+
+        # Async path (offloaded to thread)
+        async_result = await mw.abefore_agent(state, runtime)
+
+    assert sync_result is not None
+    assert async_result is not None
+    assert sync_result.keys() == async_result.keys()
+    # Both return 2 messages: reminder + user content
+    assert len(sync_result["messages"]) == 2
+    assert len(async_result["messages"]) == 2
+    # IDs match
+    assert sync_result["messages"][0].id == async_result["messages"][0].id
+    assert sync_result["messages"][1].id == async_result["messages"][1].id
+
+
+async def test_abefore_agent_returns_none_on_timeout() -> None:
+    """If _inject() exceeds the timeout, abefore_agent returns None gracefully."""
+    import time
+
+    mw = DynamicContextMiddleware()
+
+    def blocking_inject(state):
+        time.sleep(10)  # Simulate a blocking call that far exceeds the timeout
+        return {"messages": [HumanMessage(content="should not reach")]}
+
+    with (
+        mock.patch.object(mw, "_inject", blocking_inject),
+        mock.patch(
+            "deerflow.agents.middlewares.dynamic_context_middleware._INJECT_TIMEOUT_SECONDS",
+            0.1,
+        ),
+    ):
+        state = {"messages": [HumanMessage(content="Hello", id="msg-1")]}
+        runtime = SimpleNamespace(context={})
+        result = await mw.abefore_agent(state, runtime)
+
+    assert result is None

backend/tests/fixtures/replay/write_read_file.ultra.events.json

@@ -0,0 +1,132 @@
+{
+  "scenario": "write_read_file",
+  "mode": "ultra",
+  "events": [
+    {
+      "event": "metadata",
+      "keys": [
+        "run_id",
+        "thread_id"
+      ]
+    },
+    {
+      "event": "values",
+      "keys": [
+        "artifacts",
+        "messages",
+        "viewed_images"
+      ]
+    },
+    {
+      "event": "values",
+      "keys": [
+        "artifacts",
+        "messages",
+        "thread_data",
+        "viewed_images"
+      ]
+    },
+    {
+      "event": "values",
+      "keys": [
+        "artifacts",
+        "messages",
+        "thread_data",
+        "viewed_images"
+      ]
+    },
+    {
+      "event": "values",
+      "keys": [
+        "artifacts",
+        "messages",
+        "thread_data",
+        "viewed_images"
+      ]
+    },
+    {
+      "event": "values",
+      "keys": [
+        "artifacts",
+        "messages",
+        "thread_data",
+        "title",
+        "viewed_images"
+      ]
+    },
+    {
+      "event": "values",
+      "keys": [
+        "artifacts",
+        "messages",
+        "thread_data",
+        "title",
+        "viewed_images"
+      ]
+    },
+    {
+      "event": "values",
+      "keys": [
+        "artifacts",
+        "messages",
+        "thread_data",
+        "title",
+        "viewed_images"
+      ]
+    },
+    {
+      "event": "values",
+      "keys": [
+        "artifacts",
+        "messages",
+        "thread_data",
+        "title",
+        "viewed_images"
+      ]
+    },
+    {
+      "event": "values",
+      "keys": [
+        "artifacts",
+        "messages",
+        "thread_data",
+        "title",
+        "viewed_images"
+      ]
+    },
+    {
+      "event": "values",
+      "keys": [
+        "artifacts",
+        "messages",
+        "thread_data",
+        "title",
+        "viewed_images"
+      ]
+    },
+    {
+      "event": "values",
+      "keys": [
+        "artifacts",
+        "messages",
+        "thread_data",
+        "title",
+        "viewed_images"
+      ]
+    },
+    {
+      "event": "values",
+      "keys": [
+        "artifacts",
+        "messages",
+        "thread_data",
+        "title",
+        "viewed_images"
+      ]
+    },
+    {
+      "event": "end",
+      "keys": null
+    }
+  ]
+}

backend/tests/fixtures/replay/write_read_file.ultra.json

@@ -0,0 +1,243 @@
+{
+  "scenario": "write_read_file",
+  "mode": "ultra",
+  "model": "sre/gpt-5",
+  "prompt": "Using your own file tools directly, create the file /mnt/user-data/outputs/note.txt with exactly this content: hi from replay. Then read that same file back and reply with its exact contents. Do NOT delegate to a subagent and do NOT use the task tool — do it yourself. Do not ask any clarifying questions.",
+  "context": {
+    "is_bootstrap": false,
+    "mode": "ultra",
+    "thinking_enabled": true,
+    "is_plan_mode": true,
+    "subagent_enabled": true
+  },
+  "turns": [
+    {
+      "caller": "lead_agent",
+      "conversation_hash": "9c50eda6ab7e8593dabccbdeadc70a4a7bf778b2c0c3f275f1f96cf2c8ab58db",
+      "input_hash": "27aeb4c11bff2c3ebc182fe52a06556823c21928620a400c7f26be9733c31f3f",
+      "output": {
+        "type": "ai",
+        "data": {
+          "content": "",
+          "additional_kwargs": {},
+          "response_metadata": {
+            "finish_reason": "tool_calls",
+            "model_name": "sre/gpt-5",
+            "model_provider": "openai"
+          },
+          "type": "ai",
+          "name": null,
+          "id": "lc_run--019ea641-acda-7423-9a9f-79725057bc20",
+          "tool_calls": [
+            {
+              "name": "write_file",
+              "args": {
+                "description": "Create the requested output file with exact content",
+                "path": "/mnt/user-data/outputs/note.txt",
+                "content": "hi from replay."
+              },
+              "id": "call_FV7zhKonjx5CAa1RwIcKihpi",
+              "type": "tool_call"
+            }
+          ],
+          "invalid_tool_calls": [],
+          "usage_metadata": {
+            "input_tokens": 3664,
+            "output_tokens": 434,
+            "total_tokens": 4098,
+            "input_token_details": {
+              "audio": 0,
+              "cache_read": 3584
+            },
+            "output_token_details": {
+              "audio": 0,
+              "reasoning": 384
+            }
+          }
+        }
+      }
+    },
+    {
+      "caller": "middleware:title",
+      "conversation_hash": "3598aeb87e221ca8f554e4d61ce6d5e8801754606fa5c95a89c38bd6cb623045",
+      "input_hash": "75101f9faa453b1a35deff920b1e3c1a9f0b013a7627fbbaa03436752776b953",
+      "output": {
+        "type": "ai",
+        "data": {
+          "content": "Direct File Creation and Readback",
+          "additional_kwargs": {},
+          "response_metadata": {
+            "finish_reason": "stop",
+            "model_name": "sre/gpt-5",
+            "model_provider": "openai"
+          },
+          "type": "ai",
+          "name": null,
+          "id": "lc_run--019ea641-cf52-7793-900e-15ad4f032c0e",
+          "tool_calls": [],
+          "invalid_tool_calls": [],
+          "usage_metadata": {
+            "input_tokens": 104,
+            "output_tokens": 656,
+            "total_tokens": 760,
+            "input_token_details": {
+              "audio": 0,
+              "cache_read": 0
+            },
+            "output_token_details": {
+              "audio": 0,
+              "reasoning": 640
+            }
+          }
+        }
+      }
+    },
+    {
+      "caller": "lead_agent",
+      "conversation_hash": "6af134379b2a9efa01b4f63032f88211d5f38f459f8bed621eb6c65e8e05c1f9",
+      "input_hash": "f7468603a43d301fcc0167c2f7cd10e53137bfc584f1b3d776614b7a612ed7a6",
+      "output": {
+        "type": "ai",
+        "data": {
+          "content": "",
+          "additional_kwargs": {},
+          "response_metadata": {
+            "finish_reason": "tool_calls",
+            "model_name": "sre/gpt-5",
+            "model_provider": "openai"
+          },
+          "type": "ai",
+          "name": null,
+          "id": "lc_run--019ea641-f523-7d60-a416-b051fba469a2",
+          "tool_calls": [
+            {
+              "name": "read_file",
+              "args": {
+                "description": "Verify contents to echo back exactly",
+                "path": "/mnt/user-data/outputs/note.txt"
+              },
+              "id": "call_YevFCnLcjWfWHaZm8wwMpEk8",
+              "type": "tool_call"
+            }
+          ],
+          "invalid_tool_calls": [],
+          "usage_metadata": {
+            "input_tokens": 3719,
+            "output_tokens": 35,
+            "total_tokens": 3754,
+            "input_token_details": {
+              "audio": 0,
+              "cache_read": 3584
+            },
+            "output_token_details": {
+              "audio": 0,
+              "reasoning": 0
+            }
+          }
+        }
+      }
+    },
+    {
+      "caller": "lead_agent",
+      "conversation_hash": "04751c4f7b0107b78b5c97d417063883fd586f5ebcbc4acf79be6cb3c0cdaec1",
+      "input_hash": "218645dabc6926a1dbdf45dd20fba8a41e1e690cef78d7752566db3acf5a36ce",
+      "output": {
+        "type": "ai",
+        "data": {
+          "content": "hi from replay.",
+          "additional_kwargs": {},
+          "response_metadata": {
+            "finish_reason": "stop",
+            "model_name": "sre/gpt-5",
+            "model_provider": "openai"
+          },
+          "type": "ai",
+          "name": null,
+          "id": "lc_run--019ea641-ff38-7751-9c2b-cc648811883b",
+          "tool_calls": [],
+          "invalid_tool_calls": [],
+          "usage_metadata": {
+            "input_tokens": 3768,
+            "output_tokens": 8,
+            "total_tokens": 3776,
+            "input_token_details": {
+              "audio": 0,
+              "cache_read": 3584
+            },
+            "output_token_details": {
+              "audio": 0,
+              "reasoning": 0
+            }
+          }
+        }
+      }
+    },
+    {
+      "caller": "suggest_agent",
+      "conversation_hash": "8b98ebdbb53e88f000556c4753adede8eaa076ff6fd7b8a1285bfd18aee8144d",
+      "input_hash": "dcd855d389d7179a1e4bc7074fa9ba7ce697570af8947225d6bacb538f14a0cb",
+      "output": {
+        "type": "ai",
+        "data": {
+          "content": "[\n  \"Can you show the file size and last modified time of /mnt/user-data/outputs/note.txt?\",\n  \"List the contents of /mnt/user-data/outputs/ to confirm the file exists.\",\n  \"Append 'second line' to /mnt/user-data/outputs/note.txt and print its new contents.\"\n]",
+          "additional_kwargs": {
+            "refusal": null
+          },
+          "response_metadata": {
+            "token_usage": {
+              "completion_tokens": 909,
+              "prompt_tokens": 224,
+              "total_tokens": 1133,
+              "completion_tokens_details": {
+                "accepted_prediction_tokens": 0,
+                "audio_tokens": 0,
+                "reasoning_tokens": 832,
+                "rejected_prediction_tokens": 0
+              },
+              "prompt_tokens_details": {
+                "audio_tokens": 0,
+                "cached_tokens": 0
+              },
+              "latency_checkpoint": {
+                "engine_tbt_ms": 12,
+                "engine_ttft_ms": 324,
+                "engine_ttlt_ms": 10965,
+                "pre_inference_ms": 153,
+                "service_tbt_ms": 12,
+                "service_ttft_ms": 849,
+                "service_ttlt_ms": 11491,
+                "total_duration_ms": 11351,
+                "user_visible_ttft_ms": 696
+              }
+            },
+            "model_provider": "openai",
+            "model_name": "sre/gpt-5",
+            "system_fingerprint": null,
+            "id": "chatcmpl-DoPFALdwiyEDYOIN7wFYhqBrr6eTA",
+            "service_tier": "default",
+            "finish_reason": "stop",
+            "logprobs": null
+          },
+          "type": "ai",
+          "name": null,
+          "id": "lc_run--019ea642-0eac-78f1-a506-931e343184f1-0",
+          "tool_calls": [],
+          "invalid_tool_calls": [],
+          "usage_metadata": {
+            "input_tokens": 224,
+            "output_tokens": 909,
+            "total_tokens": 1133,
+            "input_token_details": {
+              "audio": 0,
+              "cache_read": 0
+            },
+            "output_token_details": {
+              "audio": 0,
+              "reasoning": 832
+            }
+          }
+        }
+      }
+    }
+  ]
+}

backend/tests/replay_provider.py

@@ -0,0 +1,384 @@
+"""Replay a recorded LLM trace deterministically — the "replay" half of
+record/replay e2e (mirrors open-design's ``mocks/`` golden traces).
+
+A fixture is a JSON file capturing the *real* model calls of one scenario,
+keyed by a normalized hash of the **caller + input** each call received::
+
+    {
+      "scenario": "write_read_file",
+      "mode": "ultra",
+      "model": "gpt-5.5",
+      "turns": [
+        {
+          "caller": "lead_agent",
+          "conversation_hash": "<sha256>",
+          "input_hash": "<sha256>",
+          "output": <message dict>,
+        },
+        ...
+      ]
+    }
+
+Why hash-by-input (not turn index)
+----------------------------------
+A real run makes model calls from several callers — the lead agent's own turns,
+``TitleMiddleware`` (auto-title), memory, and possibly subagents. They interleave
+and their count/order is not something we want a replay to depend on. Matching by
+a normalized hash of the *input messages* means each call gets back exactly the
+output that was recorded for that input, regardless of order or which middleware
+issued it. The caller name (``lead_agent``, ``middleware:title``,
+``suggest_agent``, ``subagent:*``, ...) is included so two different model
+callers with the same conversation text do not compete for the same replay
+bucket. That keeps the in-graph, deterministic title call part of the recording;
+memory/summarization, by contrast, are disabled in the replay config
+(``_replay_fixture.py``) because their background, debounced timing is not
+reproducible across runs.
+
+Volatile fields (UUID thread/run/user ids, timestamps, dates, tmp/home paths)
+are normalized out before hashing so a recording replays across processes with
+different temp dirs. The same ``hash_messages`` is used by the recorder
+(``scripts/record_gateway.py``) and here, so record and replay agree by
+construction.
+
+This lives in ``tests/`` (not in the publishable ``deerflow-harness`` package),
+matching the repo convention for test-only fakes (cf. ``FakeToolCallingModel`` in
+``_agent_e2e_helpers.py``). In-process tests get ``tests/`` on ``sys.path`` for
+free via pytest; a standalone replay gateway just needs ``PYTHONPATH`` to include
+``backend/tests`` so the config ``use:`` below resolves.
+
+Point a config model's ``use`` at this class and set the fixture via env::
+
+    models:
+      - name: replay-model
+        use: replay_provider:ReplayChatModel
+        model: gpt-5.5            # placeholder; ignored
+
+    DEERFLOW_REPLAY_FIXTURE=/path/to/write_read_file.ultra.json
+
+A cache miss raises loudly with a diagnostic — that is the signal that the
+replayed run diverged from the recording (graph changed, a new volatile field
+slipped through normalization, or a non-deterministic tool result changed a
+downstream input). Re-record or extend normalization; never pass silently.
+
+Recording lives outside production code too (``scripts/record_gateway.py`` +
+``scripts/build_fixture_from_jsonl.py``); CI consumes the fixtures through this
+replay side with no API key.
+"""
+
+from __future__ import annotations
+
+import hashlib
+import json
+import os
+import re
+from collections import deque
+from collections.abc import Iterator
+from typing import Any
+
+from langchain_core.callbacks import BaseCallbackHandler, CallbackManagerForLLMRun
+from langchain_core.language_models.chat_models import BaseChatModel
+from langchain_core.messages import AIMessage, AIMessageChunk, BaseMessage, messages_from_dict
+from langchain_core.outputs import ChatGeneration, ChatGenerationChunk, ChatResult
+from langchain_core.runnables import Runnable
+from pydantic import PrivateAttr
+
+_FIXTURE_ENV = "DEERFLOW_REPLAY_FIXTURE"
+_DEFAULT_CALLER = "lead_agent"
+_CALLER_TAG_PREFIXES = ("middleware:", "subagent:")
+_CALLER_NAME_ALIASES = {
+    # TitleMiddleware uses this run_name and tags the call as middleware:title.
+    # Some execution paths do not preserve the tag down to the model callback,
+    # so keep the run_name and tag in the same replay namespace.
+    "title_agent": "middleware:title",
+}
+
+# Process-wide record of replay misses. A miss raises inside the model, but the
+# gateway's LLMErrorHandlingMiddleware swallows it into a normal assistant error
+# message — so the SSE *event shapes* are unchanged and a shape-only golden stays
+# green on a stale fixture. The in-process Layer-1 test inspects this list to fail
+# loud on a miss instead. (Layer-2 already fails on a miss: the recorded turns
+# never render.)
+_replay_misses: list[str] = []
+
+
+def replay_misses() -> list[str]:
+    """Hashes that missed the fixture since the last reset (see ``_replay_misses``)."""
+    return list(_replay_misses)
+
+
+def reset_replay_misses() -> None:
+    _replay_misses.clear()
+
+
+def _normalize_caller(caller: str | None) -> str:
+    value = _normalize_text(str(caller or "").strip())
+    if not value:
+        return _DEFAULT_CALLER
+    return _CALLER_NAME_ALIASES.get(value, value)
+
+
+def _caller_from_tags(tags: list[str] | None) -> str | None:
+    for tag in tags or []:
+        if isinstance(tag, str) and (tag == _DEFAULT_CALLER or tag.startswith(_CALLER_TAG_PREFIXES)):
+            return tag
+    return None
+
+
+def caller_identity(*, name: str | None = None, tags: list[str] | None = None) -> str:
+    """Stable model-caller identity shared by record and replay.
+
+    Tags win because graph middleware and subagents already use them as the
+    explicit caller marker. ``run_name`` is exposed to callbacks as ``name`` and
+    covers route-level callers such as ``suggest_agent``.
+    """
+    return _normalize_caller(_caller_from_tags(tags) or name)
+
+
+# Volatile substrings that differ between a recording run and a replay run but
+# carry no semantic weight for matching. Normalized to stable placeholders
+# before hashing so the same logical input hashes identically across processes.
+# The frontend injects a per-request ``<system-reminder>`` (current date, weekday,
+# dynamic context) that the backend-direct path does not — and its date/weekday
+# change every day. Strip the whole block before hashing so a fixture replays
+# (a) across days and (b) from both the browser and direct-POST paths.
+_SYSTEM_REMINDER_RE = re.compile(r"<system-reminder>.*?</system-reminder>", re.DOTALL)
+_UUID_RE = re.compile(r"[0-9a-fA-F]{8}-[0-9a-fA-F]{4}-[0-9a-fA-F]{4}-[0-9a-fA-F]{4}-[0-9a-fA-F]{12}")
+_ISO_TS_RE = re.compile(r"\d{4}-\d{2}-\d{2}[T ]\d{2}:\d{2}:\d{2}(?:\.\d+)?(?:Z|[+-]\d{2}:?\d{2})?")
+_DATE_RE = re.compile(r"\d{4}-\d{2}-\d{2}")
+# Absolute temp/home roots used for per-run isolation (macOS + Linux + DEER_FLOW_HOME tmp).
+_PATH_RE = re.compile(r"(?:/private)?/(?:var/folders|tmp)/[^\s\"']*")
+
+
+def _normalize_text(text: str) -> str:
+    text = _SYSTEM_REMINDER_RE.sub("", text)
+    text = _UUID_RE.sub("<UUID>", text)
+    text = _ISO_TS_RE.sub("<TS>", text)
+    text = _DATE_RE.sub("<DATE>", text)
+    text = _PATH_RE.sub("<PATH>", text)
+    return text
+
+
+def _content_to_text(content: Any) -> str:
+    if isinstance(content, str):
+        return content
+    if isinstance(content, list):
+        parts: list[str] = []
+        for block in content:
+            if isinstance(block, dict):
+                parts.append(block.get("text", "") or json.dumps(block, sort_keys=True, ensure_ascii=False))
+            else:
+                parts.append(str(block))
+        return "".join(parts)
+    return str(content)
+
+
+def _canonical_messages(messages: list[BaseMessage]) -> str:
+    """Project messages to a stable shape that excludes volatile metadata/ids.
+
+    Keeps only what determines which recorded turn to replay: the conversation
+    (human / ai / tool messages — role, text content, tool-call name+args). Drops
+    ``id``, ``response_metadata``, ``usage_metadata``, ``tool_call_id`` (all
+    volatile), then normalizes embedded volatile substrings.
+
+    **The system message is excluded entirely.** The lead-agent system prompt is
+    a living, frequently-edited implementation detail (its wording changes across
+    PRs), not part of the front-back contract this harness verifies. Hashing it
+    would make every fixture go stale — and red-fail on unrelated PRs — the moment
+    anyone edits the prompt. The conversation flow (user input -> tool calls ->
+    results -> answer) is the stable key that identifies a recorded turn.
+    """
+    projected: list[dict[str, Any]] = []
+    for message in messages:
+        # Exclude the system prompt from the match key — see docstring. It is the
+        # most-edited part of the prompt and not part of the contract under test.
+        if message.type == "system":
+            continue
+        content = _normalize_text(_content_to_text(message.content))
+        tool_calls = getattr(message, "tool_calls", None)
+        # Drop messages that are empty after normalization — e.g. a turn that was
+        # nothing but a frontend-injected <system-reminder>. They carry no
+        # decision-relevant content and differ between client paths.
+        if not content.strip() and not tool_calls:
+            continue
+        entry: dict[str, Any] = {"type": message.type, "content": content}
+        if tool_calls:
+            entry["tool_calls"] = [{"name": tc.get("name"), "args": tc.get("args")} for tc in tool_calls]
+        name = getattr(message, "name", None)
+        if name:
+            entry["name"] = name
+        projected.append(entry)
+    raw = json.dumps(projected, sort_keys=True, ensure_ascii=False)
+    return _normalize_text(raw)
+
+
+def hash_messages(messages: list[BaseMessage]) -> str:
+    """Legacy stable hash of only a model call's conversation input."""
+    return hashlib.sha256(_canonical_messages(messages).encode("utf-8")).hexdigest()
+
+
+def hash_replay_input(messages: list[BaseMessage], *, caller: str | None) -> str:
+    """Stable replay key for a caller-specific model input."""
+    return hash_input_key(hash_messages(messages), caller=caller)
+
+
+def hash_input_key(conversation_hash: str, *, caller: str | None) -> str:
+    """Namespace a conversation hash by caller identity.
+
+    Keeping this as ``hash(caller + legacy_conversation_hash)`` lets existing
+    fixtures migrate without a live-model re-record: their old ``input_hash`` is
+    exactly the conversation hash.
+    """
+    payload = json.dumps(
+        {"caller": _normalize_caller(caller), "conversation_hash": conversation_hash},
+        sort_keys=True,
+        ensure_ascii=False,
+    )
+    return hashlib.sha256(payload.encode("utf-8")).hexdigest()
+
+
+def _load_fixture(fixture_path: str) -> dict[str, deque[AIMessage]]:
+    with open(fixture_path, encoding="utf-8") as handle:
+        payload = json.load(handle)
+    table: dict[str, deque[AIMessage]] = {}
+    for index, turn in enumerate(payload.get("turns", [])):
+        input_hash = turn["input_hash"]
+        (message,) = messages_from_dict([turn["output"]])
+        if not isinstance(message, AIMessage):
+            raise ValueError(f"replay fixture {fixture_path!r} turn {index} output is {type(message).__name__}, expected AIMessage")
+        table.setdefault(input_hash, deque()).append(message)
+    return table
+
+
+class ReplayChatModel(BaseChatModel):
+    """Returns the recorded assistant output whose input matches this call.
+
+    ``bind_tools`` is a no-op returning ``self`` — recorded turns already carry
+    the real ``tool_calls``, so the agent dispatches them as if a live model had
+    produced them.
+    """
+
+    _table: dict[str, deque] = PrivateAttr(default_factory=dict)
+    _fixture_path: str = PrivateAttr(default="")
+    _run_callers: dict[str, str] = PrivateAttr(default_factory=dict)
+
+    def __init__(self, **kwargs: Any) -> None:
+        # Ignore provider noise the factory forwards from config (model, api_key,
+        # base_url, ...). Fixture path comes from the ``fixture`` kwarg or env.
+        fixture_path = kwargs.pop("fixture", None) or os.environ.get(_FIXTURE_ENV)
+        callbacks = kwargs.pop("callbacks", None)
+        super().__init__(callbacks=callbacks)
+        if not fixture_path:
+            raise ValueError(f"ReplayChatModel needs a fixture path via the ``fixture`` kwarg or ${_FIXTURE_ENV}")
+        self._fixture_path = fixture_path
+        self._table = _load_fixture(fixture_path)
+        self.callbacks = [*(self.callbacks or []), _ReplayCallerCapture(self._run_callers)]
+
+    @property
+    def _llm_type(self) -> str:
+        return "deerflow-replay"
+
+    def _caller_from_run_manager(self, run_manager: CallbackManagerForLLMRun | None) -> str:
+        if run_manager is None:
+            if len(self._run_callers) == 1:
+                # Some async LangGraph paths fire on_chat_model_start with the
+                # caller metadata but invoke the model implementation without a
+                # run_manager. When there is only one pending start event, it is
+                # the current call; use it so record/replay share the same
+                # caller key.
+                return self._run_callers.pop(next(iter(self._run_callers)))
+            return _DEFAULT_CALLER
+        run_id = str(getattr(run_manager, "run_id", ""))
+        caller = self._run_callers.pop(run_id, None)
+        if caller:
+            return caller
+        return caller_identity(
+            name=getattr(run_manager, "run_name", None) or getattr(run_manager, "name", None),
+            tags=getattr(run_manager, "tags", None),
+        )
+
+    def _match(self, messages: list[BaseMessage], run_manager: CallbackManagerForLLMRun | None = None) -> AIMessage:
+        caller = self._caller_from_run_manager(run_manager)
+        key = hash_replay_input(messages, caller=caller)
+        bucket = self._table.get(key)
+        if not bucket:
+            # Backward compatibility for fixtures recorded before caller-aware
+            # keys. New recordings write caller-aware ``input_hash`` values.
+            legacy_key = hash_messages(messages)
+            bucket = self._table.get(legacy_key)
+            if bucket:
+                key = legacy_key
+        if not bucket:
+            _replay_misses.append(key)
+            preview = _canonical_messages(messages)
+            raise KeyError(
+                f"replay miss: no recorded output for input hash {key} in {self._fixture_path!r}. "
+                "The replayed run diverged from the recording (graph changed, a non-deterministic tool result "
+                "altered a downstream input, or a volatile field slipped past normalization). "
+                f"Caller: {caller!r}. "
+                f"Known hashes: {sorted(self._table)}. "
+                f"Normalized input (first 800 chars): {preview[:800]!r}"
+            )
+        return bucket.popleft()
+
+    def _generate(
+        self,
+        messages: list[BaseMessage],
+        stop: list[str] | None = None,
+        run_manager: CallbackManagerForLLMRun | None = None,
+        **kwargs: Any,
+    ) -> ChatResult:
+        return ChatResult(generations=[ChatGeneration(message=self._match(messages, run_manager))])
+
+    def _stream(
+        self,
+        messages: list[BaseMessage],
+        stop: list[str] | None = None,
+        run_manager: CallbackManagerForLLMRun | None = None,
+        **kwargs: Any,
+    ) -> Iterator[ChatGenerationChunk]:
+        turn = self._match(messages, run_manager)
+        text = turn.content if isinstance(turn.content, str) else ""
+        chunk = ChatGenerationChunk(
+            message=AIMessageChunk(
+                content=turn.content,
+                tool_calls=turn.tool_calls,
+                additional_kwargs=turn.additional_kwargs,
+                id=turn.id,
+            )
+        )
+        if run_manager is not None and text:
+            run_manager.on_llm_new_token(text, chunk=chunk)
+        yield chunk
+
+    def bind_tools(self, tools: Any, **kwargs: Any) -> Runnable:  # type: ignore[override]
+        return self
+
+
+class _ReplayCallerCapture(BaseCallbackHandler):
+    def __init__(self, run_callers: dict[str, str]) -> None:
+        self._run_callers = run_callers
+
+    def on_chat_model_start(
+        self,
+        serialized: dict,
+        messages: list[list[BaseMessage]],
+        *,
+        run_id: Any = None,
+        tags: list[str] | None = None,
+        name: str | None = None,
+        **kwargs: Any,
+    ) -> None:
+        if run_id is not None:
+            self._run_callers[str(run_id)] = caller_identity(name=name, tags=tags)
+
+
+# Re-export so the recorder shares the exact hashing logic.
+__all__ = [
+    "ReplayChatModel",
+    "caller_identity",
+    "hash_input_key",
+    "hash_messages",
+    "hash_replay_input",
+    "replay_misses",
+    "reset_replay_misses",
+]

backend/tests/seed_runs_router.py

@@ -0,0 +1,100 @@
+"""Test-only run/message seeder for the multi-run render-order e2e (issue #3352).
+
+Mounted **only** by ``scripts/run_replay_gateway.py`` (the replay e2e gateway)
+and never by the production app, so it cannot ship. It lets a Playwright spec
+stand up a thread with >=2 runs whose per-run messages exercise the frontend's
+reload / history-rebuild ordering path — with no real model, no recording, and
+no API key.
+
+Why a seeder instead of recording a conversation: issue #3352 only reproduces
+when the checkpoint no longer holds the older messages (post-compression), so
+the frontend rebuilds them from the per-run history endpoints. A seeder lets us
+create exactly that precondition deterministically — runs in the run store +
+per-run ``category="message"`` events, and **no checkpoint** — so on reload the
+buggy ``findLatestUnloadedRunIndex`` + prepend in ``core/threads/hooks.ts`` is
+the sole source of truth and its reversed order becomes observable.
+
+It writes through the gateway's OWN ``app.state.run_store`` +
+``app.state.run_event_store`` using the request's auth context, so the seeded
+``user_id`` matches the browser session that reads it back. The event shape
+mirrors exactly what ``runtime/journal.py`` writes for real runs
+(``event_type`` ``llm.human.input`` / ``llm.ai.response``, ``category``
+``"message"``, ``content`` = ``message.model_dump()``, ``metadata.caller`` =
+``"lead_agent"``).
+"""
+
+from __future__ import annotations
+
+from typing import Literal
+
+from fastapi import APIRouter, Request
+from pydantic import BaseModel
+
+router = APIRouter(prefix="/api/test-only", tags=["test-only"])
+
+# Mirror runtime/journal.py: human prompts are recorded as ``llm.human.input``
+# and assistant turns as ``llm.ai.response``; both land in ``category="message"``.
+_EVENT_TYPE = {"human": "llm.human.input", "ai": "llm.ai.response"}
+
+
+class SeedMessage(BaseModel):
+    role: Literal["human", "ai"]
+    content: str
+    id: str
+
+
+class SeedRun(BaseModel):
+    run_id: str
+    # ISO timestamp; RunManager.list_by_thread sorts newest-first by created_at,
+    # so a later created_at must mean a later run for the ordering to be faithful.
+    created_at: str
+    messages: list[SeedMessage]
+
+
+class SeedRunsBody(BaseModel):
+    thread_id: str
+    runs: list[SeedRun]
+
+
+@router.post("/seed-runs")
+async def seed_runs(body: SeedRunsBody, request: Request) -> dict:
+    """Seed runs + per-run message events for the authenticated user.
+
+    No checkpoint is written: that is the whole point — it forces the frontend's
+    reload path to rebuild history from the per-run endpoints (the #3352 bug
+    site) instead of the (correctly ordered) checkpoint snapshot.
+    """
+    from langchain_core.messages import AIMessage, HumanMessage
+
+    run_store = request.app.state.run_store
+    event_store = request.app.state.run_event_store
+
+    for run in body.runs:
+        # user_id defaults (AUTO) to the request's auth context, matching the
+        # browser session that will read these runs back via GET /runs.
+        await run_store.put(
+            run.run_id,
+            thread_id=body.thread_id,
+            assistant_id="lead_agent",
+            status="success",
+            created_at=run.created_at,
+        )
+        events = []
+        for m in run.messages:
+            msg = (HumanMessage if m.role == "human" else AIMessage)(content=m.content, id=m.id)
+            events.append(
+                {
+                    "thread_id": body.thread_id,
+                    "run_id": run.run_id,
+                    "event_type": _EVENT_TYPE[m.role],
+                    "category": "message",
+                    "content": msg.model_dump(),
+                    "metadata": {"caller": "lead_agent"},
+                    "created_at": run.created_at,
+                }
+            )
+        # One batch per run so seq is monotonic and run1's messages precede
+        # run2's; the gateway reads them back per-run anyway.
+        await event_store.put_batch(events)
+
+    return {"ok": True, "thread_id": body.thread_id, "runs": len(body.runs)}

backend/tests/test_app_config_reload.py

@@ -140,6 +140,57 @@ def test_app_config_defaults_empty_database_to_sqlite(tmp_path, monkeypatch):
     assert config.database.sqlite_dir == ".deer-flow/data"
 
 
+def test_app_config_coerces_commented_out_list_sections(tmp_path, monkeypatch):
+    """Commenting out every entry under a list key makes PyYAML parse it as None.
+
+    Regression for the documented ``cp config.example.yaml config.yaml`` flow
+    (issue #1444): such a config must load with empty lists instead of raising
+    ``Input should be a valid list``.
+    """
+    config_path = tmp_path / "config.yaml"
+    extensions_path = tmp_path / "extensions_config.json"
+    _write_extensions_config(extensions_path)
+    config_path.write_text(
+        yaml.safe_dump(
+            {
+                "sandbox": {"use": "deerflow.sandbox.local:LocalSandboxProvider"},
+                "models": None,
+                "tools": None,
+                "tool_groups": None,
+            }
+        ),
+        encoding="utf-8",
+    )
+    monkeypatch.setenv("DEER_FLOW_EXTENSIONS_CONFIG_PATH", str(extensions_path))
+
+    config = AppConfig.from_file(str(config_path))
+
+    assert config.models == []
+    assert config.tools == []
+    assert config.tool_groups == []
+
+
+def test_app_config_warns_when_no_models_configured(tmp_path, monkeypatch, caplog):
+    config_path = tmp_path / "config.yaml"
+    extensions_path = tmp_path / "extensions_config.json"
+    _write_extensions_config(extensions_path)
+    config_path.write_text(
+        yaml.safe_dump(
+            {
+                "sandbox": {"use": "deerflow.sandbox.local:LocalSandboxProvider"},
+                "models": None,
+            }
+        ),
+        encoding="utf-8",
+    )
+    monkeypatch.setenv("DEER_FLOW_EXTENSIONS_CONFIG_PATH", str(extensions_path))
+
+    with caplog.at_level("WARNING", logger="deerflow.config.app_config"):
+        AppConfig.from_file(str(config_path))
+
+    assert "No models are configured" in caplog.text
+
+
 def test_get_app_config_reloads_when_file_changes(tmp_path, monkeypatch):
     config_path = tmp_path / "config.yaml"
     extensions_path = tmp_path / "extensions_config.json"

backend/tests/test_auth_middleware.py

@@ -4,6 +4,7 @@ import pytest
 from starlette.testclient import TestClient
 
 from app.gateway.auth_middleware import AuthMiddleware, _is_public
+from app.gateway.csrf_middleware import CSRFMiddleware
 
 # ── _is_public unit tests ─────────────────────────────────────────────────
 
@@ -38,6 +39,8 @@ def test_public_paths(path: str):
         "/api/threads/123/uploads",
         "/api/agents",
         "/api/channels",
+        "/api/channels/providers",
+        "/api/channels/slack/connect",
         "/api/runs/stream",
         "/api/threads/123/runs",
         "/api/v1/auth/me",
@@ -88,7 +91,9 @@ def test_unknown_api_path_is_protected():
 
 def _make_app():
     """Create a minimal FastAPI app with AuthMiddleware for testing."""
-    from fastapi import FastAPI
+    from fastapi import FastAPI, Request
+
+    from deerflow.runtime.user_context import get_effective_user_id
 
     app = FastAPI()
     app.add_middleware(AuthMiddleware)
@@ -98,8 +103,16 @@ def _make_app():
         return {"status": "ok"}
 
     @app.get("/api/v1/auth/me")
-    async def auth_me():
-        return {"id": "1", "email": "test@test.com"}
+    async def auth_me(request: Request):
+        from app.gateway.deps import get_current_user_from_request
+
+        user = await get_current_user_from_request(request)
+        return {
+            "id": str(user.id),
+            "email": user.email,
+            "system_role": user.system_role,
+            "needs_setup": user.needs_setup,
+        }
 
     @app.get("/api/v1/auth/setup-status")
     async def setup_status():
@@ -109,6 +122,29 @@ def _make_app():
     async def models_get():
         return {"models": []}
 
+    @app.get("/api/whoami")
+    async def whoami(request: Request):
+        user = request.state.user
+        return {
+            "id": str(user.id),
+            "email": getattr(user, "email", None),
+            "system_role": getattr(user, "system_role", None),
+            "context_user_id": get_effective_user_id(),
+        }
+
+    @app.get("/api/current-user-from-dep")
+    async def current_user_from_dep(request: Request):
+        from app.gateway.deps import get_current_user_from_request
+
+        user = await get_current_user_from_request(request)
+        state_user = request.state.user
+        return {
+            "id": str(user.id),
+            "state_id": str(state_user.id),
+            "auth_source": request.state.auth_source,
+            "context_user_id": get_effective_user_id(),
+        }
+
     @app.put("/api/mcp/config")
     async def mcp_put():
         return {"ok": True}
@@ -132,8 +168,24 @@ def _make_app():
     return app
 
 
+def _make_auth_csrf_app():
+    """Create a minimal app with production middleware ordering."""
+    from fastapi import FastAPI
+
+    app = FastAPI()
+    app.add_middleware(AuthMiddleware)
+    app.add_middleware(CSRFMiddleware)
+
+    @app.post("/api/threads/abc/runs/stream")
+    async def protected_mutation():
+        return {"ok": True}
+
+    return app
+
+
 @pytest.fixture
-def client():
+def client(monkeypatch):
+    monkeypatch.delenv("DEER_FLOW_AUTH_DISABLED", raising=False)
     return TestClient(_make_app())
 
 
@@ -161,6 +213,139 @@ def test_protected_path_no_cookie_returns_401(client):
     assert body["detail"]["code"] == "not_authenticated"
 
 
+def test_auth_disabled_allows_protected_path_without_cookie(monkeypatch):
+    monkeypatch.setenv("DEER_FLOW_AUTH_DISABLED", "1")
+    client = TestClient(_make_app())
+
+    res = client.get("/api/models")
+
+    assert res.status_code == 200
+    assert res.json() == {"models": []}
+
+
+def test_auth_disabled_stamps_e2e_admin_user_without_cookie(monkeypatch):
+    monkeypatch.setenv("DEER_FLOW_AUTH_DISABLED", "1")
+    client = TestClient(_make_app())
+
+    res = client.get("/api/whoami")
+
+    assert res.status_code == 200
+    assert res.json() == {
+        "id": "e2e-user",
+        "email": "e2e@test.local",
+        "system_role": "admin",
+        "context_user_id": "e2e-user",
+    }
+
+
+def test_auth_disabled_auth_me_reuses_middleware_user_without_cookie(monkeypatch):
+    monkeypatch.setenv("DEER_FLOW_AUTH_DISABLED", "1")
+    client = TestClient(_make_app())
+
+    res = client.get("/api/v1/auth/me")
+
+    assert res.status_code == 200
+    assert res.json() == {
+        "id": "e2e-user",
+        "email": "e2e@test.local",
+        "system_role": "admin",
+        "needs_setup": False,
+    }
+
+
+def test_auth_disabled_does_not_clobber_valid_session_cookie(monkeypatch):
+    from types import SimpleNamespace
+
+    async def fake_current_user(request):
+        return SimpleNamespace(
+            id="session-user",
+            email="session@test.local",
+            system_role="user",
+            needs_setup=False,
+        )
+
+    monkeypatch.setenv("DEER_FLOW_AUTH_DISABLED", "1")
+    monkeypatch.setattr("app.gateway.deps.get_current_user_from_request", fake_current_user)
+    client = TestClient(_make_app())
+
+    res = client.get("/api/whoami", cookies={"access_token": "valid-session"})
+
+    assert res.status_code == 200
+    assert res.json() == {
+        "id": "session-user",
+        "email": "session@test.local",
+        "system_role": "user",
+        "context_user_id": "session-user",
+    }
+
+
+def test_auth_disabled_does_not_clobber_internal_auth_identity(monkeypatch):
+    from app.gateway.internal_auth import create_internal_auth_headers
+    from deerflow.runtime.user_context import DEFAULT_USER_ID
+
+    monkeypatch.setenv("DEER_FLOW_AUTH_DISABLED", "1")
+    client = TestClient(_make_app())
+
+    res = client.get(
+        "/api/current-user-from-dep",
+        headers=create_internal_auth_headers(),
+    )
+
+    assert res.status_code == 200
+    assert res.json() == {
+        "id": DEFAULT_USER_ID,
+        "state_id": DEFAULT_USER_ID,
+        "auth_source": "internal",
+        "context_user_id": DEFAULT_USER_ID,
+    }
+
+
+def test_auth_disabled_skips_csrf_for_state_changing_requests(monkeypatch):
+    monkeypatch.setenv("DEER_FLOW_AUTH_DISABLED", "1")
+    client = TestClient(_make_auth_csrf_app())
+
+    res = client.post("/api/threads/abc/runs/stream")
+
+    assert res.status_code == 200
+    assert res.json() == {"ok": True}
+
+
+def test_auth_disabled_is_ignored_in_explicit_production_env(monkeypatch):
+    monkeypatch.setenv("DEER_FLOW_AUTH_DISABLED", "1")
+    monkeypatch.setenv("DEER_FLOW_ENV", "production")
+    client = TestClient(_make_app())
+
+    res = client.get("/api/models")
+
+    assert res.status_code == 401
+
+
+def test_auth_disabled_startup_warning_when_effective(monkeypatch, caplog):
+    from app.gateway.auth_disabled import warn_if_auth_disabled_enabled
+
+    monkeypatch.setenv("DEER_FLOW_AUTH_DISABLED", "1")
+    monkeypatch.delenv("DEER_FLOW_ENV", raising=False)
+    monkeypatch.delenv("ENVIRONMENT", raising=False)
+
+    with caplog.at_level("WARNING", logger="app.gateway.auth_disabled"):
+        warn_if_auth_disabled_enabled()
+
+    assert "authentication is bypassed" in caplog.text
+    assert "e2e-user" in caplog.text
+
+
+def test_auth_disabled_startup_warning_suppressed_in_explicit_production_env(monkeypatch, caplog):
+    from app.gateway.auth_disabled import warn_if_auth_disabled_enabled
+
+    monkeypatch.setenv("DEER_FLOW_AUTH_DISABLED", "1")
+    monkeypatch.setenv("ENVIRONMENT", "production")
+
+    with caplog.at_level("WARNING", logger="app.gateway.auth_disabled"):
+        warn_if_auth_disabled_enabled()
+
+    assert "authentication is bypassed" not in caplog.text
+
+
 def test_protected_path_with_junk_cookie_rejected(client):
     """Junk cookie → 401. Middleware strictly validates the JWT now
     (AUTH_TEST_PLAN test 7.5.8); it no longer silently passes bad

backend/tests/test_channel_connections_config.py

@@ -0,0 +1,40 @@
+"""Tests for user-facing IM channel connection configuration."""
+
+from deerflow.config.channel_connections_config import ChannelConnectionsConfig
+
+
+def test_channel_connections_disabled_by_default():
+    config = ChannelConnectionsConfig()
+
+    assert config.enabled is False
+    assert config.slack.enabled is False
+    assert config.telegram.enabled is False
+    assert config.discord.enabled is False
+
+
+def test_enabled_channel_connections_do_not_require_public_url_or_encryption_key():
+    config = ChannelConnectionsConfig.model_validate(
+        {
+            "enabled": True,
+            "telegram": {
+                "enabled": True,
+                "bot_username": "deerflow_bot",
+            },
+            "slack": {"enabled": True},
+            "discord": {"enabled": True},
+        }
+    )
+
+    assert config.enabled is True
+    assert config.provider_status("telegram") == {"enabled": True, "configured": True}
+    assert config.provider_status("slack") == {"enabled": True, "configured": True}
+    assert config.provider_status("discord") == {"enabled": True, "configured": True}
+
+
+def test_provider_status_reports_disabled_and_unknown_providers():
+    config = ChannelConnectionsConfig.model_validate({"enabled": True})
+
+    assert config.provider_status("slack") == {"enabled": False, "configured": False}
+    assert config.provider_status("telegram") == {"enabled": False, "configured": False}
+    assert config.provider_status("discord") == {"enabled": False, "configured": False}
+    assert config.provider_status("unknown") == {"enabled": False, "configured": False}

backend/tests/test_channel_connections_repository.py

@@ -0,0 +1,202 @@
+"""Tests for per-user IM channel connection persistence."""
+
+from __future__ import annotations
+
+from datetime import UTC, datetime, timedelta
+
+import pytest
+from sqlalchemy import select
+
+from deerflow.persistence.channel_connections import (
+    ChannelConnectionRepository,
+    ChannelConnectionRow,
+    ChannelCredentialCipher,
+    ChannelCredentialRow,
+)
+
+
+@pytest.fixture
+async def repo(tmp_path):
+    from deerflow.persistence.engine import close_engine, get_session_factory, init_engine
+
+    url = f"sqlite+aiosqlite:///{tmp_path / 'channels.db'}"
+    await init_engine("sqlite", url=url, sqlite_dir=str(tmp_path))
+    try:
+        yield ChannelConnectionRepository(
+            get_session_factory(),
+            cipher=ChannelCredentialCipher.from_key("test-encryption-key"),
+        )
+    finally:
+        await close_engine()
+
+
+class TestChannelConnectionRepository:
+    @pytest.mark.anyio
+    async def test_connections_are_listed_per_owner(self, repo):
+        alice = await repo.upsert_connection(
+            owner_user_id="alice",
+            provider="slack",
+            external_account_id="U-alice",
+            external_account_name="Alice",
+            workspace_id="T1",
+            workspace_name="Team One",
+            scopes=["chat:write"],
+        )
+        await repo.upsert_connection(
+            owner_user_id="bob",
+            provider="slack",
+            external_account_id="U-bob",
+            external_account_name="Bob",
+            workspace_id="T1",
+            workspace_name="Team One",
+            scopes=["chat:write"],
+        )
+
+        results = await repo.list_connections("alice")
+
+        assert [item["id"] for item in results] == [alice["id"]]
+        assert results[0]["owner_user_id"] == "alice"
+        assert results[0]["provider"] == "slack"
+        assert results[0]["scopes"] == ["chat:write"]
+        assert "encrypted_access_token" not in results[0]
+
+    @pytest.mark.anyio
+    async def test_upsert_connection_updates_existing_provider_identity(self, repo):
+        first = await repo.upsert_connection(
+            owner_user_id="alice",
+            provider="telegram",
+            external_account_id="42",
+            external_account_name="Alice",
+            workspace_id=None,
+            workspace_name=None,
+            status="pending",
+        )
+        second = await repo.upsert_connection(
+            owner_user_id="alice",
+            provider="telegram",
+            external_account_id="42",
+            external_account_name="Alice Telegram",
+            workspace_id=None,
+            workspace_name=None,
+            status="connected",
+        )
+
+        assert second["id"] == first["id"]
+        assert second["status"] == "connected"
+        assert second["external_account_name"] == "Alice Telegram"
+        assert len(await repo.list_connections("alice")) == 1
+
+    @pytest.mark.anyio
+    async def test_credentials_are_encrypted_at_rest_and_decrypted_by_repository(self, repo):
+        connection = await repo.upsert_connection(
+            owner_user_id="alice",
+            provider="slack",
+            external_account_id="U-alice",
+            workspace_id="T1",
+        )
+        expires_at = datetime.now(UTC) + timedelta(hours=1)
+
+        await repo.store_credentials(
+            connection["id"],
+            access_token="xoxb-secret-access-token",
+            refresh_token="secret-refresh-token",
+            token_type="Bearer",
+            expires_at=expires_at,
+            extra={"bot_user_id": "B123"},
+        )
+
+        async with repo.session_factory() as session:
+            row = (await session.execute(select(ChannelCredentialRow))).scalar_one()
+            assert row.encrypted_access_token is not None
+            assert "xoxb-secret-access-token" not in row.encrypted_access_token
+            assert "secret-refresh-token" not in (row.encrypted_refresh_token or "")
+            assert "B123" not in (row.encrypted_extra_json or "")
+
+        credentials = await repo.get_credentials(connection["id"])
+
+        assert credentials is not None
+        assert credentials["access_token"] == "xoxb-secret-access-token"
+        assert credentials["refresh_token"] == "secret-refresh-token"
+        assert credentials["token_type"] == "Bearer"
+        assert credentials["expires_at"] == expires_at
+        assert credentials["extra"] == {"bot_user_id": "B123"}
+
+    @pytest.mark.anyio
+    async def test_conversations_are_scoped_by_connection(self, repo):
+        alice = await repo.upsert_connection(
+            owner_user_id="alice",
+            provider="slack",
+            external_account_id="U-alice",
+            workspace_id="T1",
+        )
+        bob = await repo.upsert_connection(
+            owner_user_id="bob",
+            provider="slack",
+            external_account_id="U-bob",
+            workspace_id="T1",
+        )
+
+        await repo.set_thread_id(
+            connection_id=alice["id"],
+            owner_user_id="alice",
+            provider="slack",
+            external_conversation_id="C-shared",
+            external_topic_id="1710000000.000100",
+            thread_id="thread-alice",
+        )
+        await repo.set_thread_id(
+            connection_id=bob["id"],
+            owner_user_id="bob",
+            provider="slack",
+            external_conversation_id="C-shared",
+            external_topic_id="1710000000.000100",
+            thread_id="thread-bob",
+        )
+
+        assert await repo.get_thread_id(alice["id"], "C-shared", "1710000000.000100") == "thread-alice"
+        assert await repo.get_thread_id(bob["id"], "C-shared", "1710000000.000100") == "thread-bob"
+
+    @pytest.mark.anyio
+    async def test_disconnect_connection_revokes_owner_connection_and_removes_credentials(self, repo):
+        connection = await repo.upsert_connection(
+            owner_user_id="alice",
+            provider="telegram",
+            external_account_id="42",
+        )
+        await repo.store_credentials(connection["id"], access_token="secret-token")
+
+        disconnected = await repo.disconnect_connection(
+            connection_id=connection["id"],
+            owner_user_id="alice",
+        )
+
+        assert disconnected is True
+        async with repo.session_factory() as session:
+            connection_row = await session.get(ChannelConnectionRow, connection["id"])
+            credential_row = await session.get(ChannelCredentialRow, connection["id"])
+        assert connection_row is not None
+        assert connection_row.status == "revoked"
+        assert credential_row is None
+        assert (
+            await repo.find_connection_by_external_identity(
+                provider="telegram",
+                external_account_id="42",
+            )
+            is None
+        )
+
+    @pytest.mark.anyio
+    async def test_disconnect_connection_is_owner_scoped(self, repo):
+        connection = await repo.upsert_connection(
+            owner_user_id="alice",
+            provider="telegram",
+            external_account_id="42",
+        )
+
+        disconnected = await repo.disconnect_connection(
+            connection_id=connection["id"],
+            owner_user_id="bob",
+        )
+
+        assert disconnected is False
+        assert (await repo.list_connections("alice"))[0]["status"] == "connected"

backend/tests/test_channel_connections_router.py

@@ -0,0 +1,287 @@
+"""Router tests for browser-connectable IM channels."""
+
+from __future__ import annotations
+
+from uuid import UUID
+
+from _router_auth_helpers import make_authed_test_app
+from fastapi.testclient import TestClient
+
+from app.gateway.auth.models import User
+from app.gateway.routers import channel_connections
+from deerflow.config.channel_connections_config import ChannelConnectionsConfig
+
+
+def _user() -> User:
+    return User(
+        id=UUID("11111111-2222-3333-4444-555555555555"),
+        email="alice@example.com",
+        password_hash="x",
+        system_role="user",
+    )
+
+
+async def _make_repo(tmp_path):
+    from deerflow.persistence.channel_connections import ChannelConnectionRepository
+    from deerflow.persistence.engine import get_session_factory, init_engine
+
+    await init_engine("sqlite", url=f"sqlite+aiosqlite:///{tmp_path / 'router.db'}", sqlite_dir=str(tmp_path))
+    return ChannelConnectionRepository(get_session_factory())
+
+
+def _make_app(config: ChannelConnectionsConfig, repo, channels_config: dict | None = None):
+    app = make_authed_test_app(user_factory=_user)
+    app.state.channel_connections_config = config
+    app.state.channel_connection_repo = repo
+    app.state.channels_config = channels_config or {}
+    app.include_router(channel_connections.router)
+    return app
+
+
+def _enabled_connections_config() -> ChannelConnectionsConfig:
+    return ChannelConnectionsConfig.model_validate(
+        {
+            "enabled": True,
+            "telegram": {"enabled": True, "bot_username": "deerflow_bot"},
+            "slack": {"enabled": True},
+            "discord": {"enabled": True},
+        }
+    )
+
+
+def _channels_config() -> dict:
+    return {
+        "telegram": {"enabled": True, "bot_token": "telegram-token"},
+        "slack": {"enabled": True, "bot_token": "xoxb-operator", "app_token": "xapp-operator"},
+        "discord": {"enabled": True, "bot_token": "discord-bot"},
+    }
+
+
+def test_get_providers_uses_existing_channels_config(tmp_path):
+    import anyio
+
+    repo = anyio.run(_make_repo, tmp_path)
+    app = _make_app(_enabled_connections_config(), repo, _channels_config())
+
+    with TestClient(app) as client:
+        response = client.get("/api/channels/providers")
+
+    assert response.status_code == 200
+    body = response.json()
+    assert body["enabled"] is True
+    by_provider = {item["provider"]: item for item in body["providers"]}
+    assert by_provider["telegram"]["configured"] is True
+    assert by_provider["telegram"]["auth_mode"] == "deep_link"
+    assert by_provider["slack"]["configured"] is True
+    assert by_provider["slack"]["auth_mode"] == "binding_code"
+    assert by_provider["discord"]["configured"] is True
+    assert by_provider["discord"]["auth_mode"] == "binding_code"
+
+    anyio.run(repo.close)
+
+
+def test_get_providers_reports_unconfigured_when_runtime_channel_is_missing(tmp_path):
+    import anyio
+
+    repo = anyio.run(_make_repo, tmp_path)
+    app = _make_app(_enabled_connections_config(), repo, {"telegram": {"enabled": True, "bot_token": "telegram-token"}})
+
+    with TestClient(app) as client:
+        response = client.get("/api/channels/providers")
+
+    assert response.status_code == 200
+    by_provider = {item["provider"]: item for item in response.json()["providers"]}
+    assert by_provider["telegram"]["configured"] is True
+    assert by_provider["slack"]["configured"] is False
+    assert by_provider["slack"]["connectable"] is False
+    assert "channels.slack" in by_provider["slack"]["unavailable_reason"]
+    assert by_provider["discord"]["configured"] is False
+    assert "channels.discord" in by_provider["discord"]["unavailable_reason"]
+
+    anyio.run(repo.close)
+
+
+def test_get_connections_returns_current_user_connections_only(tmp_path):
+    import anyio
+
+    repo = anyio.run(_make_repo, tmp_path)
+
+    async def seed_connections():
+        await repo.upsert_connection(
+            owner_user_id=str(_user().id),
+            provider="telegram",
+            external_account_id="42",
+            external_account_name="Alice",
+            status="connected",
+        )
+        await repo.upsert_connection(
+            owner_user_id="other-user",
+            provider="telegram",
+            external_account_id="99",
+            external_account_name="Bob",
+            status="connected",
+        )
+
+    anyio.run(seed_connections)
+    app = _make_app(_enabled_connections_config(), repo, _channels_config())
+
+    with TestClient(app) as client:
+        response = client.get("/api/channels/connections")
+
+    assert response.status_code == 200
+    body = response.json()
+    assert len(body["connections"]) == 1
+    assert body["connections"][0]["provider"] == "telegram"
+    assert body["connections"][0]["external_account_id"] == "42"
+
+    anyio.run(repo.close)
+
+
+def test_connect_telegram_returns_deep_link_and_persists_state(tmp_path):
+    import anyio
+
+    repo = anyio.run(_make_repo, tmp_path)
+    app = _make_app(_enabled_connections_config(), repo, _channels_config())
+
+    with TestClient(app) as client:
+        response = client.post("/api/channels/telegram/connect")
+
+    assert response.status_code == 200
+    body = response.json()
+    assert body["provider"] == "telegram"
+    assert body["mode"] == "deep_link"
+    assert body["url"].startswith("https://t.me/deerflow_bot?start=")
+    assert body["code"]
+    assert "/start" in body["instruction"]
+
+    async def count_states():
+        return await repo.count_oauth_states(owner_user_id=str(_user().id), provider="telegram")
+
+    assert anyio.run(count_states) == 1
+
+    anyio.run(repo.close)
+
+
+def test_connect_slack_returns_binding_command_and_persists_state(tmp_path):
+    import anyio
+
+    repo = anyio.run(_make_repo, tmp_path)
+    app = _make_app(_enabled_connections_config(), repo, _channels_config())
+
+    with TestClient(app) as client:
+        response = client.post("/api/channels/slack/connect")
+
+    assert response.status_code == 200
+    body = response.json()
+    assert body["provider"] == "slack"
+    assert body["mode"] == "binding_code"
+    assert body["url"] is None
+    assert body["code"]
+    assert body["instruction"] == f"Send /connect {body['code']} to the DeerFlow Slack bot."
+
+    async def count_states():
+        return await repo.count_oauth_states(owner_user_id=str(_user().id), provider="slack")
+
+    assert anyio.run(count_states) == 1
+
+    anyio.run(repo.close)
+
+
+def test_connect_discord_returns_binding_command_and_persists_state(tmp_path):
+    import anyio
+
+    repo = anyio.run(_make_repo, tmp_path)
+    app = _make_app(_enabled_connections_config(), repo, _channels_config())
+
+    with TestClient(app) as client:
+        response = client.post("/api/channels/discord/connect")
+
+    assert response.status_code == 200
+    body = response.json()
+    assert body["provider"] == "discord"
+    assert body["mode"] == "binding_code"
+    assert body["url"] is None
+    assert body["code"]
+    assert body["instruction"] == f"Send /connect {body['code']} to the DeerFlow Discord bot."
+
+    async def count_states():
+        return await repo.count_oauth_states(owner_user_id=str(_user().id), provider="discord")
+
+    assert anyio.run(count_states) == 1
+
+    anyio.run(repo.close)
+
+
+def test_connect_unconfigured_runtime_channel_returns_400(tmp_path):
+    import anyio
+
+    repo = anyio.run(_make_repo, tmp_path)
+    app = _make_app(_enabled_connections_config(), repo, {})
+
+    with TestClient(app) as client:
+        response = client.post("/api/channels/slack/connect")
+
+    assert response.status_code == 400
+    assert "channels.slack" in response.json()["detail"]
+
+    anyio.run(repo.close)
+
+
+def test_disconnect_connection_revokes_current_user_connection(tmp_path):
+    import anyio
+
+    repo = anyio.run(_make_repo, tmp_path)
+
+    async def seed_connection():
+        connection = await repo.upsert_connection(
+            owner_user_id=str(_user().id),
+            provider="telegram",
+            external_account_id="42",
+            status="connected",
+        )
+        return connection["id"]
+
+    connection_id = anyio.run(seed_connection)
+    app = _make_app(_enabled_connections_config(), repo, _channels_config())
+
+    with TestClient(app) as client:
+        response = client.delete(f"/api/channels/connections/{connection_id}")
+
+    assert response.status_code == 204
+
+    async def get_connection_status():
+        return (await repo.list_connections(str(_user().id)))[0]["status"]
+
+    assert anyio.run(get_connection_status) == "revoked"
+
+    anyio.run(repo.close)
+
+
+def test_disconnect_connection_is_current_user_scoped(tmp_path):
+    import anyio
+
+    repo = anyio.run(_make_repo, tmp_path)
+
+    async def seed_connection():
+        connection = await repo.upsert_connection(
+            owner_user_id="other-user",
+            provider="telegram",
+            external_account_id="42",
+            status="connected",
+        )
+        return connection["id"]
+
+    connection_id = anyio.run(seed_connection)
+    app = _make_app(_enabled_connections_config(), repo, _channels_config())
+
+    with TestClient(app) as client:
+        response = client.delete(f"/api/channels/connections/{connection_id}")
+
+    assert response.status_code == 404
+
+    async def get_connection_status():
+        return (await repo.list_connections("other-user"))[0]["status"]
+
+    assert anyio.run(get_connection_status) == "connected"
+
+    anyio.run(repo.close)

backend/tests/test_checkpointer.py

@@ -2,7 +2,9 @@
 
 import sys
 import tomllib
+from concurrent.futures import ThreadPoolExecutor
 from pathlib import Path
+from threading import Barrier, Event, Lock
 from unittest.mock import AsyncMock, MagicMock, patch
 
 import pytest
@@ -10,12 +12,14 @@ import pytest
 import deerflow.config.app_config as app_config_module
 from deerflow.config.checkpointer_config import (
     CheckpointerConfig,
+    ensure_config_loaded,
     get_checkpointer_config,
     load_checkpointer_config_from_dict,
     set_checkpointer_config,
 )
 from deerflow.runtime.checkpointer import get_checkpointer, reset_checkpointer
 from deerflow.runtime.checkpointer.provider import POSTGRES_INSTALL
+from deerflow.runtime.store import get_store, reset_store
 from deerflow.runtime.store.provider import POSTGRES_STORE_INSTALL
 
 
@@ -25,10 +29,90 @@ def reset_state():
     app_config_module._app_config = None
     set_checkpointer_config(None)
     reset_checkpointer()
+    reset_store()
     yield
     app_config_module._app_config = None
     set_checkpointer_config(None)
     reset_checkpointer()
+    reset_store()
+
+
+class _BlockingSingletonContext:
+    def __init__(self, value: object, entered: Event, release: Event, stats: dict[str, object]):
+        self._value = value
+        self._entered = entered
+        self._release = release
+        self._stats = stats
+
+    def __enter__(self):
+        with self._stats["lock"]:
+            self._stats["enters"] += 1
+            self._entered.set()
+        assert self._release.wait(timeout=3), "timed out waiting to release singleton initialization"
+        return self._value
+
+    def __exit__(self, exc_type, exc, tb):
+        with self._stats["lock"]:
+            self._stats["exits"] += 1
+        return False
+
+
+class _BlockingSingletonFactory:
+    def __init__(self):
+        self.value = object()
+        self.entered = Event()
+        self.release = Event()
+        self.stats = {"enters": 0, "exits": 0, "lock": Lock()}
+
+    def context_manager(self, _config):
+        return _BlockingSingletonContext(self.value, self.entered, self.release, self.stats)
+
+    def enter_count(self) -> int:
+        with self.stats["lock"]:
+            return self.stats["enters"]
+
+    def exit_count(self) -> int:
+        with self.stats["lock"]:
+            return self.stats["exits"]
+
+
+class _TrackingLock:
+    def __init__(self):
+        self._lock = Lock()
+        self.acquired = Event()
+
+    def acquire(self, *args, **kwargs):
+        acquired = self._lock.acquire(*args, **kwargs)
+        if acquired:
+            self.acquired.set()
+        return acquired
+
+    def release(self):
+        self._lock.release()
+
+    def __enter__(self):
+        self.acquire()
+        return self
+
+    def __exit__(self, exc_type, exc, tb):
+        self.release()
+        return False
+
+    def locked(self) -> bool:
+        return self._lock.locked()
+
+
+def _call_getter_concurrently(getter, workers: int = 8) -> list[object]:
+    ready = Barrier(workers + 1)
+
+    def worker():
+        ready.wait(timeout=3)
+        return getter()
+
+    with ThreadPoolExecutor(max_workers=workers) as executor:
+        futures = [executor.submit(worker) for _ in range(workers)]
+        ready.wait(timeout=3)
+        return [future.result(timeout=3) for future in futures]
 
 
 # ---------------------------------------------------------------------------
@@ -67,6 +151,26 @@ class TestCheckpointerConfig:
         set_checkpointer_config(None)
         assert get_checkpointer_config() is None
 
+    def test_ensure_config_loaded_loads_app_config_when_uninitialized(self):
+        def fake_get_app_config():
+            load_checkpointer_config_from_dict({"type": "memory"})
+
+        with patch("deerflow.config.app_config.get_app_config", side_effect=fake_get_app_config) as mock_get_app_config:
+            ensure_config_loaded()
+
+        mock_get_app_config.assert_called_once()
+        config = get_checkpointer_config()
+        assert config is not None
+        assert config.type == "memory"
+
+    def test_ensure_config_loaded_skips_explicit_config(self):
+        load_checkpointer_config_from_dict({"type": "memory"})
+
+        with patch("deerflow.config.app_config.get_app_config") as mock_get_app_config:
+            ensure_config_loaded()
+
+        mock_get_app_config.assert_not_called()
+
     def test_invalid_type_raises(self):
         with pytest.raises(Exception):
             load_checkpointer_config_from_dict({"type": "unknown"})
@@ -118,7 +222,7 @@ class TestGetCheckpointer:
         """get_checkpointer should return InMemorySaver when not configured."""
         from langgraph.checkpoint.memory import InMemorySaver
 
-        with patch("deerflow.runtime.checkpointer.provider.get_app_config", side_effect=FileNotFoundError):
+        with patch("deerflow.config.app_config.get_app_config", side_effect=FileNotFoundError):
             cp = get_checkpointer()
         assert cp is not None
         assert isinstance(cp, InMemorySaver)
@@ -287,6 +391,143 @@ class TestGetCheckpointer:
         mock_saver_instance.setup.assert_called_once()
 
 
+class TestSyncSingletonThreadSafety:
+    def test_store_reset_clears_singleton(self):
+        load_checkpointer_config_from_dict({"type": "memory"})
+        store1 = get_store()
+        reset_store()
+        store2 = get_store()
+        assert store1 is not store2
+
+    def test_concurrent_checkpointer_getter_creates_one_instance(self):
+        load_checkpointer_config_from_dict({"type": "memory"})
+        factory = _BlockingSingletonFactory()
+
+        with patch("deerflow.runtime.checkpointer.provider._sync_checkpointer_cm", side_effect=factory.context_manager):
+            futures_started = ThreadPoolExecutor(max_workers=1)
+            try:
+                result_future = futures_started.submit(_call_getter_concurrently, get_checkpointer)
+                assert factory.entered.wait(timeout=3)
+                factory.release.wait(timeout=0.05)
+                factory.release.set()
+                results = result_future.result(timeout=3)
+            finally:
+                futures_started.shutdown(wait=True)
+
+        assert all(result is factory.value for result in results)
+        assert factory.enter_count() == 1
+
+    def test_concurrent_store_getter_creates_one_instance(self):
+        load_checkpointer_config_from_dict({"type": "memory"})
+        factory = _BlockingSingletonFactory()
+
+        with patch("deerflow.runtime.store.provider._sync_store_cm", side_effect=factory.context_manager):
+            futures_started = ThreadPoolExecutor(max_workers=1)
+            try:
+                result_future = futures_started.submit(_call_getter_concurrently, get_store)
+                assert factory.entered.wait(timeout=3)
+                factory.release.wait(timeout=0.05)
+                factory.release.set()
+                results = result_future.result(timeout=3)
+            finally:
+                futures_started.shutdown(wait=True)
+
+        assert all(result is factory.value for result in results)
+        assert factory.enter_count() == 1
+
+    def test_checkpointer_loads_config_outside_singleton_lock(self):
+        tracking_lock = _TrackingLock()
+
+        def fake_ensure_config_loaded():
+            assert not tracking_lock.locked()
+            load_checkpointer_config_from_dict({"type": "memory"})
+
+        with (
+            patch("deerflow.runtime.checkpointer.provider._checkpointer_lock", tracking_lock),
+            patch("deerflow.runtime.checkpointer.provider.ensure_config_loaded", side_effect=fake_ensure_config_loaded),
+        ):
+            checkpointer = get_checkpointer()
+
+        assert checkpointer is not None
+        assert tracking_lock.acquired.is_set()
+
+    def test_store_loads_config_outside_singleton_lock(self):
+        tracking_lock = _TrackingLock()
+
+        def fake_ensure_config_loaded():
+            assert not tracking_lock.locked()
+            load_checkpointer_config_from_dict({"type": "memory"})
+
+        with (
+            patch("deerflow.runtime.store.provider._store_lock", tracking_lock),
+            patch("deerflow.runtime.store.provider.ensure_config_loaded", side_effect=fake_ensure_config_loaded),
+        ):
+            store = get_store()
+
+        assert store is not None
+        assert tracking_lock.acquired.is_set()
+
+    def test_checkpointer_reset_waits_for_initialization(self):
+        load_checkpointer_config_from_dict({"type": "memory"})
+        factory = _BlockingSingletonFactory()
+
+        with (
+            patch("deerflow.runtime.checkpointer.provider._sync_checkpointer_cm", side_effect=factory.context_manager),
+            ThreadPoolExecutor(max_workers=2) as executor,
+        ):
+            get_future = executor.submit(get_checkpointer)
+            assert factory.entered.wait(timeout=3)
+
+            reset_started = Event()
+
+            def reset_worker():
+                reset_started.set()
+                reset_checkpointer()
+
+            reset_future = executor.submit(reset_worker)
+            assert reset_started.wait(timeout=3)
+            factory.release.wait(timeout=0.05)
+
+            assert not reset_future.done()
+            assert factory.exit_count() == 0
+
+            factory.release.set()
+            assert get_future.result(timeout=3) is factory.value
+            reset_future.result(timeout=3)
+
+        assert factory.exit_count() == 1
+
+    def test_store_reset_waits_for_initialization(self):
+        load_checkpointer_config_from_dict({"type": "memory"})
+        factory = _BlockingSingletonFactory()
+
+        with (
+            patch("deerflow.runtime.store.provider._sync_store_cm", side_effect=factory.context_manager),
+            ThreadPoolExecutor(max_workers=2) as executor,
+        ):
+            get_future = executor.submit(get_store)
+            assert factory.entered.wait(timeout=3)
+
+            reset_started = Event()
+
+            def reset_worker():
+                reset_started.set()
+                reset_store()
+
+            reset_future = executor.submit(reset_worker)
+            assert reset_started.wait(timeout=3)
+            factory.release.wait(timeout=0.05)
+
+            assert not reset_future.done()
+            assert factory.exit_count() == 0
+
+            factory.release.set()
+            assert get_future.result(timeout=3) is factory.value
+            reset_future.result(timeout=3)
+
+        assert factory.exit_count() == 1
+
+
 class TestAsyncCheckpointer:
     @pytest.mark.anyio
     async def test_sqlite_creates_parent_dir_via_to_thread(self):
@@ -506,7 +747,7 @@ class TestClientCheckpointerFallback:
             patch("deerflow.client.get_app_config", return_value=config_mock),
             patch("deerflow.client.create_agent", side_effect=fake_create_agent),
             patch("deerflow.client.create_chat_model", return_value=MagicMock()),
-            patch("deerflow.client._build_middlewares", return_value=[]),
+            patch("deerflow.client.build_middlewares", return_value=[]),
             patch("deerflow.client.apply_prompt_template", return_value=""),
             patch("deerflow.client.DeerFlowClient._get_tools", return_value=[]),
         ):
@@ -540,7 +781,7 @@ class TestClientCheckpointerFallback:
             patch("deerflow.client.get_app_config", return_value=config_mock),
             patch("deerflow.client.create_agent", side_effect=fake_create_agent),
             patch("deerflow.client.create_chat_model", return_value=MagicMock()),
-            patch("deerflow.client._build_middlewares", return_value=[]),
+            patch("deerflow.client.build_middlewares", return_value=[]),
             patch("deerflow.client.apply_prompt_template", return_value=""),
             patch("deerflow.client.DeerFlowClient._get_tools", return_value=[]),
         ):

backend/tests/test_client.py

@@ -910,7 +910,7 @@ class TestEnsureAgent:
         with (
             patch("deerflow.client.create_chat_model"),
             patch("deerflow.client.create_agent", return_value=mock_agent),
-            patch("deerflow.client._build_middlewares", return_value=[]) as mock_build_middlewares,
+            patch("deerflow.client.build_middlewares", return_value=[]) as mock_build_middlewares,
             patch("deerflow.client.apply_prompt_template", return_value="prompt") as mock_apply_prompt,
             patch.object(client, "_get_tools", return_value=[]),
             patch("deerflow.runtime.checkpointer.get_checkpointer", return_value=MagicMock()),
@@ -935,7 +935,7 @@ class TestEnsureAgent:
         with (
             patch("deerflow.client.create_chat_model"),
             patch("deerflow.client.create_agent", return_value=mock_agent) as mock_create_agent,
-            patch("deerflow.client._build_middlewares", return_value=[]),
+            patch("deerflow.client.build_middlewares", return_value=[]),
             patch("deerflow.client.apply_prompt_template", return_value="prompt"),
             patch.object(client, "_get_tools", return_value=[]),
             patch("deerflow.runtime.checkpointer.get_checkpointer", return_value=mock_checkpointer),
@@ -960,7 +960,7 @@ class TestEnsureAgent:
         with (
             patch("deerflow.client.create_chat_model"),
             patch("deerflow.client.create_agent", return_value=mock_agent) as mock_create_agent,
-            patch("deerflow.client._build_middlewares", side_effect=fake_build_middlewares),
+            patch("deerflow.client.build_middlewares", side_effect=fake_build_middlewares),
             patch("deerflow.client.apply_prompt_template", return_value="prompt"),
             patch.object(client, "_get_tools", return_value=[]),
             patch("deerflow.runtime.checkpointer.get_checkpointer", return_value=MagicMock()),
@@ -979,7 +979,7 @@ class TestEnsureAgent:
         with (
             patch("deerflow.client.create_chat_model"),
             patch("deerflow.client.create_agent", return_value=mock_agent) as mock_create_agent,
-            patch("deerflow.client._build_middlewares", return_value=[]),
+            patch("deerflow.client.build_middlewares", return_value=[]),
             patch("deerflow.client.apply_prompt_template", return_value="prompt"),
             patch.object(client, "_get_tools", return_value=[]),
             patch("deerflow.runtime.checkpointer.get_checkpointer", return_value=None),
@@ -1957,7 +1957,7 @@ class TestScenarioAgentRecreation:
         with (
             patch("deerflow.client.create_chat_model"),
             patch("deerflow.client.create_agent", side_effect=fake_create_agent),
-            patch("deerflow.client._build_middlewares", return_value=[]),
+            patch("deerflow.client.build_middlewares", return_value=[]),
             patch("deerflow.client.apply_prompt_template", return_value="prompt"),
             patch.object(client, "_get_tools", return_value=[]),
             patch("deerflow.runtime.checkpointer.get_checkpointer", return_value=MagicMock()),
@@ -1985,7 +1985,7 @@ class TestScenarioAgentRecreation:
         with (
             patch("deerflow.client.create_chat_model"),
             patch("deerflow.client.create_agent", side_effect=fake_create_agent),
-            patch("deerflow.client._build_middlewares", return_value=[]),
+            patch("deerflow.client.build_middlewares", return_value=[]),
             patch("deerflow.client.apply_prompt_template", return_value="prompt"),
             patch.object(client, "_get_tools", return_value=[]),
             patch("deerflow.runtime.checkpointer.get_checkpointer", return_value=MagicMock()),
@@ -2010,7 +2010,7 @@ class TestScenarioAgentRecreation:
         with (
             patch("deerflow.client.create_chat_model"),
             patch("deerflow.client.create_agent", side_effect=fake_create_agent),
-            patch("deerflow.client._build_middlewares", return_value=[]),
+            patch("deerflow.client.build_middlewares", return_value=[]),
             patch("deerflow.client.apply_prompt_template", return_value="prompt"),
             patch.object(client, "_get_tools", return_value=[]),
             patch("deerflow.runtime.checkpointer.get_checkpointer", return_value=MagicMock()),

backend/tests/test_client_e2e.py

@@ -144,14 +144,14 @@ def e2e_env(tmp_path, monkeypatch):
     #    non-determinism and cost to E2E tests (title generation is already
     #    disabled via TitleConfig above, but the middleware still participates
     #    in the chain and can interfere with event ordering).
-    from deerflow.agents.lead_agent.agent import _build_middlewares as _original_build_middlewares
+    from deerflow.agents.lead_agent.agent import build_middlewares as _original_build_middlewares
     from deerflow.agents.middlewares.title_middleware import TitleMiddleware
 
     def _sync_safe_build_middlewares(*args, **kwargs):
         mws = _original_build_middlewares(*args, **kwargs)
         return [m for m in mws if not isinstance(m, TitleMiddleware)]
 
-    monkeypatch.setattr("deerflow.client._build_middlewares", _sync_safe_build_middlewares)
+    monkeypatch.setattr("deerflow.client.build_middlewares", _sync_safe_build_middlewares)
 
     return {"tmp_path": tmp_path}
 

backend/tests/test_compose_default_workers.py

@@ -0,0 +1,45 @@
+"""Regression test for the Docker Compose default Gateway worker count.
+
+The Gateway holds run state (RunManager and the stream bridge) in process, so
+the default deployment must run a single Uvicorn worker. Running more than one
+worker without a shared cross-worker stream bridge breaks run cancellation, SSE
+reconnects, request de-duplication, and IM channels (nginx has no sticky
+sessions, so requests scatter across workers that each keep their own run
+state). This test pins the safe default so it cannot silently regress to a
+multi-worker default, while still allowing operators to override it once a
+shared stream bridge exists.
+"""
+
+from __future__ import annotations
+
+import re
+from pathlib import Path
+
+import yaml
+
+REPO_ROOT = Path(__file__).resolve().parents[2]
+COMPOSE_PATH = REPO_ROOT / "docker" / "docker-compose.yaml"
+
+
+def _gateway_command() -> str:
+    """Return the gateway service command as a single string."""
+    compose = yaml.safe_load(COMPOSE_PATH.read_text(encoding="utf-8"))
+    command = compose["services"]["gateway"]["command"]
+    # ``command`` may load as a scalar string or a list depending on YAML style.
+    if isinstance(command, list):
+        command = " ".join(str(part) for part in command)
+    return command
+
+
+def test_gateway_defaults_to_single_worker():
+    """With GATEWAY_WORKERS unset, the worker count must default to 1."""
+    command = _gateway_command()
+    match = re.search(r"GATEWAY_WORKERS:-(\d+)", command)
+    assert match is not None, f"gateway command must set a GATEWAY_WORKERS default; got: {command}"
+    assert match.group(1) == "1", f"default Gateway worker count must be 1, got {match.group(1)}"
+
+
+def test_gateway_worker_count_remains_overridable():
+    """The worker count must stay configurable, not hard-coded to 1."""
+    command = _gateway_command()
+    assert "${GATEWAY_WORKERS:-1}" in command, f"worker count must use ${{GATEWAY_WORKERS:-1}} so operators can override it; got: {command}"

backend/tests/test_csrf_middleware.py

@@ -233,3 +233,15 @@ def test_non_auth_mutation_rejects_mismatched_double_submit_token():
 
     assert response.status_code == 403
     assert response.json()["detail"] == "CSRF token mismatch."
+
+
+def test_channel_posts_require_double_submit_csrf():
+    client = TestClient(_make_app(), base_url="https://deerflow.example")
+
+    response = client.post(
+        "/api/channels/slack/connect",
+        headers={"Origin": "https://deerflow.example"},
+    )
+
+    assert response.status_code == 403
+    assert response.json()["detail"] == "CSRF token missing. Include X-CSRF-Token header."

backend/tests/test_ddg_search_tools.py

@@ -0,0 +1,75 @@
+"""Unit tests for the DDGS community web search tool."""
+
+import json
+import sys
+from types import SimpleNamespace
+from unittest.mock import MagicMock, patch
+
+from deerflow.community.ddg_search import tools
+
+
+def test_resolve_ddgs_region_maps_worldwide_chinese_query_for_wikipedia() -> None:
+    assert tools._resolve_ddgs_region("\u4e16\u754c\u676f\u65b0\u95fb 2026", "wt-wt", "auto") == "cn-zh"
+
+
+def test_resolve_ddgs_region_uses_english_fallback_for_worldwide_query() -> None:
+    assert tools._resolve_ddgs_region("latest world cup news", "wt-wt", "auto") == "us-en"
+
+
+def test_resolve_ddgs_region_preserves_worldwide_for_non_wikipedia_backend() -> None:
+    assert tools._resolve_ddgs_region("latest world cup news", "wt-wt", "duckduckgo") == "wt-wt"
+
+
+def test_resolve_ddgs_region_maps_common_ddg_locale_aliases() -> None:
+    assert tools._resolve_ddgs_region("\u65e5\u672c \u30cb\u30e5\u30fc\u30b9", "jp-jp", "auto") == "jp-ja"
+    assert tools._resolve_ddgs_region("\ud55c\uad6d \ub274\uc2a4", "kr-kr", "auto") == "kr-ko"
+    assert tools._resolve_ddgs_region("\u53f0\u7063\u65b0\u805e", "tw-tzh", "auto") == "tw-zh"
+
+
+def test_search_text_passes_wikipedia_safe_region_to_ddgs(monkeypatch) -> None:
+    calls = {}
+
+    class FakeDDGS:
+        def __init__(self, timeout: int) -> None:
+            calls["timeout"] = timeout
+
+        def text(self, query: str, **kwargs):
+            calls["query"] = query
+            calls.update(kwargs)
+            return [{"title": "Result", "href": "https://example.com", "body": "Snippet"}]
+
+    monkeypatch.setitem(sys.modules, "ddgs", SimpleNamespace(DDGS=FakeDDGS))
+
+    results = tools._search_text("\u4e16\u754c\u676f\u65b0\u95fb 2026", backend="auto")
+
+    assert results == [{"title": "Result", "href": "https://example.com", "body": "Snippet"}]
+    assert calls["timeout"] == 30
+    assert calls["region"] == "cn-zh"
+    assert calls["backend"] == "auto"
+
+
+def test_web_search_tool_reads_ddgs_options_from_config() -> None:
+    with patch("deerflow.community.ddg_search.tools.get_app_config") as mock_config:
+        tool_config = MagicMock()
+        tool_config.model_extra = {
+            "max_results": 3,
+            "region": "us-en",
+            "safesearch": "off",
+            "backend": "auto",
+        }
+        mock_config.return_value.get_tool_config.return_value = tool_config
+
+        with patch("deerflow.community.ddg_search.tools._search_text") as mock_search:
+            mock_search.return_value = [{"title": "Result", "href": "https://example.com", "body": "Snippet"}]
+
+            result = tools.web_search_tool.invoke({"query": "latest news", "max_results": 8})
+            parsed = json.loads(result)
+
+    assert parsed["total_results"] == 1
+    mock_search.assert_called_once_with(
+        query="latest news",
+        max_results=3,
+        region="us-en",
+        safesearch="off",
+        backend="auto",
+    )

backend/tests/test_deferred_tool_crosscontext.py

@@ -22,7 +22,7 @@ from langchain_core.tools import tool as as_tool
 from deerflow.agents.middlewares.deferred_tool_filter_middleware import DeferredToolFilterMiddleware
 from deerflow.skills.tool_policy import filter_tools_by_skill_allowed_tools
 from deerflow.skills.types import Skill
-from deerflow.tools.builtins.tool_search import DeferredToolSetup, build_deferred_tool_setup
+from deerflow.tools.builtins.tool_search import DeferredToolSetup, assemble_deferred_tools, build_deferred_tool_setup
 from deerflow.tools.mcp_metadata import tag_mcp_tool
 
 
@@ -93,17 +93,15 @@ def test_policy_excluded_mcp_tool_not_in_catalog():
 def test_fail_closed_when_mcp_survives_without_setup(monkeypatch):
     """Finding 2: simulate a wiring regression and assert it fails loudly.
 
-    ``_assemble_deferred`` lazy-imports ``build_deferred_tool_setup`` from the
-    source module, so patch it there (not on the agent module).
+    ``assemble_deferred_tools`` references ``build_deferred_tool_setup`` as a
+    module global, so patch it in ``tool_search`` (its home module).
     """
-    from deerflow.agents.lead_agent import agent as agentmod
-
     monkeypatch.setattr(
         "deerflow.tools.builtins.tool_search.build_deferred_tool_setup",
         lambda tools, *, enabled: DeferredToolSetup(None, frozenset(), None),
     )
     with pytest.raises(RuntimeError, match="fail-closed"):
-        agentmod._assemble_deferred([tag_mcp_tool(mcp_secret)], enabled=True)
+        assemble_deferred_tools([tag_mcp_tool(mcp_secret)], enabled=True)
 
 
 def test_subagent_reentry_does_not_touch_lead_state():
@@ -146,12 +144,10 @@ def _make_skill(allowed_tools):
 
 def test_policy_denied_mcp_yields_no_tool_search_end_to_end():
     """An allowlist that denies the MCP tool gates it end-to-end: after the real
-    policy filter no MCP tool survives, so ``_assemble_deferred`` adds no
+    policy filter no MCP tool survives, so ``assemble_deferred_tools`` adds no
     tool_search (and does not fail-closed, because no MCP tool leaked through)."""
-    from deerflow.agents.lead_agent import agent as agentmod
-
     filtered = filter_tools_by_skill_allowed_tools([active_tool, tag_mcp_tool(mcp_secret)], [_make_skill(["active_tool"])])
-    final_tools, setup = agentmod._assemble_deferred(filtered, enabled=True)
+    final_tools, setup = assemble_deferred_tools(filtered, enabled=True)
 
     assert [t.name for t in final_tools] == ["active_tool"]
     assert "tool_search" not in {t.name for t in final_tools}
@@ -167,11 +163,9 @@ def test_tool_search_appended_after_policy_but_never_exposes_denied_tool():
     is derived from the already policy-filtered list — so it can never expose a
     tool the allowlist denied. Locks that contract so the ordering cannot regress.
     """
-    from deerflow.agents.lead_agent import agent as agentmod
-
     allowed = ["active_tool", "mcp_secret"]  # permits the MCP tool, does NOT list tool_search
     filtered = filter_tools_by_skill_allowed_tools([active_tool, tag_mcp_tool(mcp_secret)], [_make_skill(allowed)])
-    final_tools, setup = agentmod._assemble_deferred(filtered, enabled=True)
+    final_tools, setup = assemble_deferred_tools(filtered, enabled=True)
 
     names = {t.name for t in final_tools}
     assert "tool_search" in names  # appended despite not being in the allowlist

backend/tests/test_dev_entrypoint.py

@@ -40,6 +40,20 @@ def test_entrypoint_script_exists_and_is_posix_sh():
     assert proc.returncode == 0, proc.stderr
 
 
+def test_entrypoint_excludes_runtime_state_from_uvicorn_reload():
+    content = ENTRYPOINT.read_text(encoding="utf-8")
+
+    assert ': "${DEER_FLOW_HOME:=/app/backend/.deer-flow}"' in content
+    # sandbox must be created too, not just .deer-flow (#3459 / #3454).
+    assert 'mkdir -p "$DEER_FLOW_HOME" /app/backend/.deer-flow /app/backend/sandbox' in content
+    assert "--reload-include='*.yaml .env'" not in content
+    assert "--reload-include='*.yaml'" in content
+    assert "--reload-include='.env'" in content
+    assert "--reload-exclude=/app/backend/sandbox" in content
+    assert '--reload-exclude="$DEER_FLOW_HOME"' in content
+    assert "--reload-exclude=/app/backend/.deer-flow" in content
+
+
 def test_no_uv_extras_yields_empty_flags():
     proc = _run(None)
     assert proc.returncode == 0

backend/tests/test_discord_channel.py

@@ -2,9 +2,13 @@
 
 from __future__ import annotations
 
+from types import SimpleNamespace
+
+import pytest
+
 from app.channels.discord import DiscordChannel
 from app.channels.manager import CHANNEL_CAPABILITIES
-from app.channels.message_bus import MessageBus
+from app.channels.message_bus import InboundMessageType, MessageBus
 from app.channels.service import _CHANNEL_REGISTRY
 
 
@@ -21,3 +25,64 @@ def test_discord_channel_init() -> None:
     channel = DiscordChannel(bus=bus, config={"bot_token": "token"})
 
     assert channel.name == "discord"
+
+
+def _make_discord_message(text: str):
+    return SimpleNamespace(
+        id=111,
+        content=text,
+        author=SimpleNamespace(id=123, bot=False, display_name="alice"),
+        guild=SimpleNamespace(id=321),
+        channel=SimpleNamespace(id=456),
+        add_reaction=lambda _emoji: None,
+    )
+
+
+@pytest.mark.asyncio
+async def test_discord_bot_mention_slash_skill_routes_as_chat() -> None:
+    bus = MessageBus()
+    channel = DiscordChannel(bus=bus, config={"bot_token": "token"})
+    captured = []
+    channel._running = True
+    channel._client = SimpleNamespace(user=SimpleNamespace(id=999, mention="<@999>"))
+    channel._discord_module = SimpleNamespace(Thread=type("FakeThread", (), {}))
+    channel._publish = captured.append
+
+    async def noop(*_args, **_kwargs):
+        return None
+
+    channel._start_typing = noop
+    channel._add_reaction = noop
+
+    await channel._on_message(_make_discord_message("<@999> /data-analysis analyze uploads/foo.csv"))
+
+    assert len(captured) == 1
+    inbound = captured[0]
+    assert inbound.text == "/data-analysis analyze uploads/foo.csv"
+    assert inbound.msg_type == InboundMessageType.CHAT
+    assert inbound.topic_id == "456"
+
+
+@pytest.mark.asyncio
+async def test_discord_bot_mention_known_command_routes_as_command() -> None:
+    bus = MessageBus()
+    channel = DiscordChannel(bus=bus, config={"bot_token": "token"})
+    captured = []
+    channel._running = True
+    channel._client = SimpleNamespace(user=SimpleNamespace(id=999, mention="<@999>"))
+    channel._discord_module = SimpleNamespace(Thread=type("FakeThread", (), {}))
+    channel._publish = captured.append
+
+    async def noop(*_args, **_kwargs):
+        return None
+
+    channel._start_typing = noop
+    channel._add_reaction = noop
+
+    await channel._on_message(_make_discord_message("<@999> /help"))
+
+    assert len(captured) == 1
+    inbound = captured[0]
+    assert inbound.text == "/help"
+    assert inbound.msg_type == InboundMessageType.COMMAND
+    assert inbound.topic_id == "456"

backend/tests/test_discord_channel_connections.py

@@ -0,0 +1,88 @@
+"""Discord connection routing tests."""
+
+from __future__ import annotations
+
+from datetime import UTC, datetime, timedelta
+from unittest.mock import AsyncMock, MagicMock
+
+import pytest
+
+from app.channels.discord import DiscordChannel
+from app.channels.message_bus import InboundMessage, MessageBus
+
+
+@pytest.fixture
+async def repo(tmp_path):
+    from deerflow.persistence.channel_connections import ChannelConnectionRepository, ChannelCredentialCipher
+    from deerflow.persistence.engine import close_engine, get_session_factory, init_engine
+
+    await init_engine("sqlite", url=f"sqlite+aiosqlite:///{tmp_path / 'discord.db'}", sqlite_dir=str(tmp_path))
+    try:
+        yield ChannelConnectionRepository(
+            get_session_factory(),
+            cipher=ChannelCredentialCipher.from_key("discord-secret"),
+        )
+    finally:
+        await close_engine()
+
+
+@pytest.mark.anyio
+async def test_discord_inbound_attaches_owner_identity_from_user_level_connection(repo):
+    connection = await repo.upsert_connection(
+        owner_user_id="alice",
+        provider="discord",
+        external_account_id="987",
+        external_account_name="Alice",
+        status="connected",
+    )
+    channel = DiscordChannel(
+        bus=MessageBus(),
+        config={"bot_token": "discord-bot", "connection_repo": repo},
+    )
+    inbound = InboundMessage(
+        channel_name="discord",
+        chat_id="C123",
+        user_id="987",
+        text="hello",
+    )
+
+    attached = await channel._attach_connection_identity(inbound, guild_id="G123")
+
+    assert attached.connection_id == connection["id"]
+    assert attached.owner_user_id == "alice"
+    assert attached.workspace_id is None
+
+
+@pytest.mark.anyio
+async def test_discord_connect_command_binds_gateway_identity(repo):
+    state = "discord-bind-code"
+    await repo.create_oauth_state(
+        owner_user_id="deerflow-user-1",
+        provider="discord",
+        state=state,
+        expires_at=datetime.now(UTC) + timedelta(minutes=5),
+    )
+    channel = DiscordChannel(
+        bus=MessageBus(),
+        config={"bot_token": "discord-bot", "connection_repo": repo},
+    )
+    message = MagicMock()
+    message.author.id = 987
+    message.author.display_name = "Alice"
+    message.guild.id = 123
+    message.guild.name = "Deer Guild"
+    message.channel.id = 456
+    message.channel.send = AsyncMock()
+
+    handled = await channel._bind_connection_from_connect_code(message, state)
+
+    connections = await repo.list_connections("deerflow-user-1")
+    assert handled is True
+    assert len(connections) == 1
+    assert connections[0]["provider"] == "discord"
+    assert connections[0]["external_account_id"] == "987"
+    assert connections[0]["external_account_name"] == "Alice"
+    assert connections[0]["workspace_id"] == "123"
+    assert connections[0]["workspace_name"] == "Deer Guild"
+    assert connections[0]["metadata"]["channel_id"] == "456"
+    message.channel.send.assert_awaited_once()

backend/tests/test_gateway_runtime_cleanup.py

@@ -43,6 +43,21 @@ def test_service_launchers_always_use_gateway_runtime():
         assert "LANGGRAPH_REWRITE" not in content, path
 
 
+def test_local_dev_gateway_reload_excludes_runtime_state_with_absolute_dirs():
+    serve_sh = _read("scripts/serve.sh")
+
+    assert 'export DEER_FLOW_PROJECT_ROOT="$REPO_ROOT"' in serve_sh
+    assert 'BACKEND_RUNTIME_HOME="$REPO_ROOT/backend/.deer-flow"' in serve_sh
+    assert 'export DEER_FLOW_HOME="$BACKEND_RUNTIME_HOME"' in serve_sh
+    # Every absolute reload-exclude must be pre-created, including backend/sandbox
+    # (#3459 / #3454) — see test_uvicorn_reload_exclude.py for the mechanism.
+    assert 'mkdir -p "$DEER_FLOW_HOME" "$BACKEND_RUNTIME_HOME" "$REPO_ROOT/backend/sandbox"' in serve_sh
+    assert "--reload-exclude='$DEER_FLOW_HOME'" in serve_sh
+    assert "--reload-exclude='$BACKEND_RUNTIME_HOME'" in serve_sh
+    assert "--reload-exclude='sandbox/'" not in serve_sh
+    assert "--reload-exclude='.deer-flow/'" not in serve_sh
+
+
 def test_backend_container_only_exposes_gateway_port():
     dockerfile = _read("backend/Dockerfile")