furyhawk/ai_agent

mirror of https://github.com/furyhawk/ai_agent.git synced 2026-07-20 09:45:44 +00:00

Files

T

furyhawk 8351e73d39 feat: add Zustand stores for conversation, file preview, sidebar, theme, and knowledge base selection

- Implemented `conversation-store` for managing conversations and messages.
- Created `file-preview-store` to handle file preview state.
- Added `sidebar-store` for sidebar visibility management.
- Developed `theme-store` for theme persistence and management.
- Introduced `kb-selection-store` for managing active knowledge base selections with persistence.

chore: define API and chat types

- Added types for API responses, authentication, chat messages, conversations, and projects.
- Defined interfaces for various entities including users, sessions, and message ratings.

build: configure TypeScript and testing setup

- Set up `tsconfig.json` for TypeScript configuration.
- Created `vitest.config.ts` for testing configuration with Vitest.
- Added `vitest.setup.ts` for global test setup including mocks for Next.js router and media queries.
- Configured Vercel deployment settings in `vercel.json`.

2026-06-11 16:54:43 +08:00

2.9 KiB

Raw Permalink Blame History

Manual setup steps for ai_agent

The generator created the code. These are the one-time external setup steps that can't be automated — accounts to create, keys to copy, services to provision.

Skip ahead to "After every deploy" at the bottom for things you'll re-do regularly. Items above are one-time per environment.

Secrets

cp backend/.env.example backend/.env

Then in backend/.env:

SECRET_KEY — replace with a fresh value: openssl rand -hex 32
API_KEY — replace with a fresh value: openssl rand -hex 32

These are used to sign JWTs and authenticate service-to-service calls. Rotate at every environment promotion (dev → staging → prod each get their own).

PostgreSQL

Provision a PostgreSQL ≥ 14 instance (local: docker compose up -d db; managed: Neon / Supabase / RDS / Cloud SQL).
Set DATABASE_URL in .env to the async connection string: postgresql+asyncpg://user:pass@host:5432/dbname.
Run migrations: cd backend && uv run alembic upgrade head.

OpenAI

Create API key at https://platform.openai.com/api-keys.
Set OPENAI_API_KEY in .env.
(Optional) Set spending limit on OpenAI dashboard to avoid surprise bills.

Google OAuth

Go to https://console.cloud.google.com/ → APIs & Services → Credentials → Create OAuth client ID.
Application type: Web application.
Authorized redirect URIs: http://localhost:3033/auth/callback. Add prod URL when deploying.
Copy Client ID + Client secret → set GOOGLE_OAUTH_CLIENT_ID + GOOGLE_OAUTH_CLIENT_SECRET in .env.

RAG (milvus)

Local: docker compose up -d milvus etcd minio (already in docker-compose.yml).
Cloud: provision via Zilliz Cloud, set MILVUS_URI + MILVUS_TOKEN.
(Optional) Ingest seed documents: uv run ai_agent rag-ingest /path/to/file.pdf --collection docs.

Redis

Local: docker compose up -d redis (already in compose file).
Managed: Upstash / Redis Cloud / ElastiCache. Set REDIS_URL in .env.

Logfire (Pydantic observability)

Create account at https://logfire.pydantic.dev.
Run uv run logfire auth once locally to bootstrap.
Get write token → set LOGFIRE_TOKEN in .env for non-local environments.

After every deploy

Run database migrations: alembic upgrade head (CI step or post-deploy job).
Smoke test /api/v1/health returns {"status": "ok"}.
Frontend loads, login → dashboard flow works.
Logs flowing to your aggregator.

Where to find more

ENV_VARS.md — exhaustive env var reference
docs/deploy.md — platform-specific deployment recipes
SECURITY.md — security model + production hardening checklist
CONTRIBUTING.md — dev environment setup
docs/architecture.md — codebase layered architecture rules