Quick Start¶
Prerequisites¶
| Tool | Version |
|---|---|
| Node.js | 20+ |
| pnpm | 10+ |
| Docker + docker compose | Latest stable |
Installation and Startup¶
Step 1: Install dependencies¶
Step 2: Start infrastructure¶
Start PostgreSQL and Redis (Docker Compose for development):
Step 3: Start the backend¶
Listens on http://127.0.0.1:8787 by default. The embedded worker starts automatically.
Step 4: Start the debug console¶
Open http://localhost:5174.
Verify It Works¶
After startup, check:
- Backend logs show the runtime mode (embedded worker / api-only)
- Browser opens
http://localhost:5174 - Send a message in the console. The run should move from
queuedto executing.
Tip
If the backend is not at the default address, set the proxy target:
Single Workspace Mode¶
To serve a single workspace without a config file, point directly at a workspace path:
pnpm dev:server -- \
--workspace /absolute/path/to/workspace \
--model-dir /absolute/path/to/models \
--default-model openai-default
Optional flags: --workspace-kind chat, --tool-dir, --skill-dir, --host, --port
Info
In single workspace mode, the debug console enters the workspace automatically.
Common Commands¶
| Command | Purpose |
|---|---|
pnpm install |
Install dependencies |
pnpm infra:up |
Start PostgreSQL + Redis |
pnpm infra:down |
Stop infrastructure |
pnpm dev:server -- --config ./server.example.yaml |
Start backend (embedded worker) |
pnpm dev:server -- --api-only --config ./server.example.yaml |
Start API only |
pnpm dev:worker -- --config ./server.example.yaml |
Start standalone worker |
pnpm dev:web |
Start debug console |
pnpm build |
Full build |
pnpm test |
Run tests |
mkdocs serve |
Preview docs locally |
Next Steps¶
- Architecture Overview — Understand the system structure
- Workspace Guide — Configure agents, skills, and tools
- Deploy and Run — Unified local vs split production deployment
- Design Overview — Core design decisions