test(release): harden QA and live validation (#108589)

extensions/qa-lab/src/scenario-catalog.test.ts

@@ -623,7 +623,7 @@ describe("qa scenario catalog", () => {
     const scenario = readQaScenarioById("luna-thinking-visibility-switch");
     const config = readQaScenarioExecutionConfig("luna-thinking-visibility-switch") as
       | {
-          requiredProvider?: string;
+          liveProvider?: string;
           requiredModel?: string;
           offDirective?: string;
           maxDirective?: string;
@@ -632,7 +632,7 @@ describe("qa scenario catalog", () => {
       | undefined;
 
     expect(scenario.sourcePath).toBe("qa/scenarios/models/luna-thinking-visibility-switch.yaml");
-    expect(config?.requiredProvider).toBe("openai");
+    expect(config?.liveProvider).toBe("openai");
     expect(config?.requiredModel).toBe("gpt-5.6-luna");
     expect(config?.offDirective).toBe("/think off");
     expect(config?.maxDirective).toBe("/think medium");
@@ -939,6 +939,19 @@ describe("qa scenario catalog", () => {
     expect(subagentFanout.execution.suiteIsolation).toBe("isolated");
   });
 
+  it("settles subagent completions before reading the SQLite session store", () => {
+    const scenario = requireFlowScenario(readQaScenarioById("subagent-fanout-synthesis"));
+    const flow = JSON.stringify(scenario.execution.flow);
+    const completionWaits = [...flow.matchAll(/expectedChildCompletionMarkers/gu)].map(
+      (match) => match.index,
+    );
+    const storeReads = [...flow.matchAll(/readRawQaSessionStore/gu)].map((match) => match.index);
+
+    expect(completionWaits).toHaveLength(2);
+    expect(storeReads).toHaveLength(2);
+    expect(completionWaits.every((wait, index) => wait < (storeReads[index] ?? -1))).toBe(true);
+  });
+
   it("adds a dreaming shadow trial report scenario", () => {
     const scenario = readQaScenarioById("dreaming-shadow-trial-report");
     const config = readQaScenarioExecutionConfig("dreaming-shadow-trial-report") as

extensions/qa-lab/src/suite-runtime-flow.test.ts

@@ -60,7 +60,6 @@ const hasDiscoveryLabels = vi.hoisted(() => vi.fn());
 const reportsDiscoveryScopeLeak = vi.hoisted(() => vi.fn());
 const reportsMissingDiscoveryFiles = vi.hoisted(() => vi.fn());
 const hasModelSwitchContinuitySignal = vi.hoisted(() => vi.fn());
-const qaChannelPlugin = vi.hoisted(() => ({ id: "qa-channel" }));
 const scanGatewayLogSentinels = vi.hoisted(() => vi.fn());
 const assertNoGatewayLogSentinels = vi.hoisted(() => vi.fn());
 
@@ -159,10 +158,6 @@ vi.mock("./model-switch-eval.js", () => ({
   hasModelSwitchContinuitySignal,
 }));
 
-vi.mock("./runtime-api.js", () => ({
-  qaChannelPlugin,
-}));
-
 vi.mock("./gateway-log-sentinel.js", () => ({
   scanGatewayLogSentinels,
   assertNoGatewayLogSentinels,
@@ -296,7 +291,6 @@ describe("qa suite runtime flow", () => {
           envArg: typeof env,
           configArg: Record<string, unknown>,
         ) => Promise<unknown>;
-        qaChannelPlugin: typeof qaChannelPlugin;
         webOpenPage: (params: { url: string }) => Promise<unknown>;
       };
       constants: {
@@ -330,7 +324,6 @@ describe("qa suite runtime flow", () => {
         ensureImageGenerationConfigured,
       },
     );
-    expect(call.deps.qaChannelPlugin).toBe(qaChannelPlugin);
     expect(call.constants).toEqual({
       imageUnderstandingPngBase64: "small",
       imageUnderstandingLargePngBase64: "large",

qa/scenarios/agents/subagent-fanout-synthesis.yaml

@@ -111,6 +111,23 @@ flow:
                                   expr: "config.expectedReplyGroups.every((group) => group.some((needle) => normalizeLowercaseStringOrEmpty(text).includes(needle)))"
                               - expr: "30000"
                               - expr: "env.providerMode === 'mock-openai' ? 100 : 250"
+                          - if:
+                              # The parent reply can arrive before child transcript writes finish.
+                              # Wait before opening the store so its integrity check sees settled FTS state.
+                              expr: "Boolean(env.mock) && env.gateway.runtimeEnv.OPENCLAW_QA_FORCE_RUNTIME !== 'codex'"
+                              then:
+                                - forEach:
+                                    items:
+                                      expr: "config.expectedChildCompletionMarkers"
+                                    item: childCompletionMarker
+                                    actions:
+                                      - call: waitForOutboundMessage
+                                        args:
+                                          - ref: state
+                                          - lambda:
+                                              params: [candidate]
+                                              expr: "String(candidate.text ?? '').trim() === childCompletionMarker"
+                                          - 30000
                           - if:
                               expr: "Boolean(env.mock)"
                               then:
@@ -163,6 +180,21 @@ flow:
                           - if:
                               expr: "/timed out after/i.test(formatErrorMessage(attemptError))"
                               then:
+                                - if:
+                                    expr: "Boolean(env.mock) && env.gateway.runtimeEnv.OPENCLAW_QA_FORCE_RUNTIME !== 'codex'"
+                                    then:
+                                      - forEach:
+                                          items:
+                                            expr: "config.expectedChildCompletionMarkers"
+                                          item: childCompletionMarker
+                                          actions:
+                                            - call: waitForOutboundMessage
+                                              args:
+                                                - ref: state
+                                                - lambda:
+                                                    params: [candidate]
+                                                    expr: "String(candidate.text ?? '').trim() === childCompletionMarker"
+                                                - 30000
                                 - call: readRawQaSessionStore
                                   saveAs: timeoutStore
                                   args:
@@ -240,23 +272,4 @@ flow:
             expr: "lastError === '__done__'"
             message:
               expr: "lastError instanceof Error ? formatErrorMessage(lastError) : String(lastError ?? 'fanout retry exhausted')"
-        - if:
-            # Codex completes child sessions through its app-server path but
-            # does not relay the child marker back onto the parent QA channel.
-            # The shared assertions above already prove both child tool calls
-            # and child session rows; keep this transport-only proof OpenClaw-specific.
-            expr: "Boolean(env.mock) && env.gateway.runtimeEnv.OPENCLAW_QA_FORCE_RUNTIME !== 'codex'"
-            then:
-              - forEach:
-                  items:
-                    expr: "config.expectedChildCompletionMarkers"
-                  item: childCompletionMarker
-                  actions:
-                    - call: waitForOutboundMessage
-                      args:
-                        - ref: state
-                        - lambda:
-                            params: [candidate]
-                            expr: "String(candidate.text ?? '').trim() === childCompletionMarker"
-                        - 30000
       detailsExpr: "details"

qa/scenarios/models/luna-thinking-visibility-switch.yaml

@@ -29,7 +29,7 @@ scenario:
     kind: flow
     summary: Toggle reasoning display and GPT-5.6 Luna thinking between off/none and medium, then verify visible reasoning only on the medium turn.
     config:
-      requiredProvider: openai
+      liveProvider: openai
       requiredModel: gpt-5.6-luna
       offDirective: /think off
       maxDirective: /think medium
@@ -57,7 +57,7 @@ flow:
           value:
             expr: splitModelRef(env.primaryModel)
         - assert:
-            expr: "env.providerMode !== 'live-frontier' || (selected?.provider === config.requiredProvider && selected?.model === config.requiredModel)"
+            expr: "env.providerMode !== 'live-frontier' || (selected?.provider === config.liveProvider && selected?.model === config.requiredModel)"
             message:
               expr: "`expected live GPT-5.6 Luna, got ${env.primaryModel}`"
         - sendInbound:

src/agents/zai.live.test.ts

@@ -35,14 +35,21 @@ async function expectModelReturnsAssistantText(
     contextWindow: modelId === "glm-5.2" ? 1_000_000 : 202_800,
     maxTokens: modelId === "glm-5.2" ? 131_072 : 131_100,
   };
-  const res = await completeSimple(
-    model,
-    {
-      messages: createSingleUserPromptMessage(),
-    },
-    { apiKey: ZAI_KEY, maxTokens: 64 },
-  );
-  const text = extractNonEmptyAssistantText(res.content);
+  const runProbe = (maxTokens: number) =>
+    completeSimple(
+      model,
+      {
+        messages: createSingleUserPromptMessage(),
+      },
+      { apiKey: ZAI_KEY, maxTokens },
+    );
+  const res = await runProbe(64);
+  let text = extractNonEmptyAssistantText(res.content);
+  // GLM can spend a tiny cap entirely on hidden reasoning. Retry only a
+  // provider-declared truncation; other empty responses remain failures.
+  if (text.length === 0 && res.stopReason === "length") {
+    text = extractNonEmptyAssistantText((await runProbe(256)).content);
+  }
   expect(text.length).toBeGreaterThan(0);
 }
 

src/plugins/compat/registry.test.ts

@@ -33,7 +33,7 @@ const publicSdkContractNarrowingTiers = [
   {
     name: "bundled-only public export",
     codeSuffix: "-public-demotion",
-    count: 158,
+    count: 157,
     replacement:
       "subpath becomes internal (private-local-only); no external successor — no known external consumers",
     releaseNote: /public export.*module stays available for bundled plugins.*private-local-only/u,