addon-langgraph-agent（插件 langgraph Agent）

Add-on: LangGraph Agent

Use this skill when a project needs multi-step agent execution (tool use, checkpoints, and bounded loops) through LangGraph.

Compatibility

• Works with architect-python-uv-fastapi-sqlalchemy and architect-nextjs-bun-app.
• Can be combined with addon-langchain-llm for shared provider setup.
• Can be combined with addon-google-agent-dev-kit when ADK interop is needed.

Inputs

Collect:

• AGENT_SCOPE: tasks the agent is explicitly allowed to perform.
• CHECKPOINT_BACKEND: memory | postgres | redis.
• ALLOW_TOOLS: explicit tool allow-list.
• MAX_STEPS: default 10.
• TIMEOUT_SECONDS: default 45.

Integration Workflow

1. Add dependencies:

• Python:

uv add langgraph langchain-core pydantic-settings

• Next.js:

# Use the project's package manager (examples):
bun add @langchain/langgraph zod
pnpm add @langchain/langgraph zod

1. Add files by architecture:

• Python API:

src/{{MODULE_NAME}}/agent/state.py
src/{{MODULE_NAME}}/agent/graph.py
src/{{MODULE_NAME}}/agent/tools.py
src/{{MODULE_NAME}}/api/routes/agent.py

• Next.js:

src/lib/agent/state.ts
src/lib/agent/graph.ts
src/lib/agent/tools.ts
src/app/api/agent/runs/route.ts

1. Expose run lifecycle endpoints:

• Create run (POST /v1/agent/runs or app-equivalent route).
• Fetch run status (GET /v1/agent/runs/{run_id} or app-equivalent route).

1. Persist run telemetry:

• run_id, status, steps, started_at, finished_at, error.

Required Template

Agent run response shape

{
  "runId": "uuid",
  "status": "queued"
}

Guardrails

• Documentation contract for generated code:

- Python: write module docstrings and docstrings for public classes, methods, and functions. - Next.js/TypeScript: write JSDoc for exported components, hooks, utilities, and route handlers. - Add concise rationale comments only for non-obvious logic, invariants, or safety constraints. - Apply this contract even when using template snippets below; expand templates as needed.

• Restrict tools to explicit allow-list only.
• Enforce max-step and timeout boundaries.
• Persist failure state for post-run inspection.
• Return deterministic fallback on provider/tool failure.

Validation Checklist

• Confirm generated code includes required docstrings/JSDoc and rationale comments for non-obvious logic.

uv run ruff check . || true
uv run mypy src || true
# Use the project's package manager (examples):
bun run lint || true
pnpm run lint || true
rg -n "runId|MAX_STEPS|timeout|allow-list" src || true

• Manual checks:
• Run creation returns stable run id format.
• Agent aborts safely after max steps.

Decision Justification Rule

• Every non-trivial decision must include a concrete justification.
• Capture the alternatives considered and why they were rejected.
• State tradeoffs and residual risks for the chosen option.
• If justification is missing, treat the task as incomplete and surface it as a blocker.