Getting Started¶

Installation¶

pip install fast-agent-stack

For extras:

pip install "fast-agent-stack[anthropic,auth-jwt,db-postgres]"

Create a Project¶

mkdir myproject && cd myproject
uv venv && source .venv/bin/activate
uv pip install fast-agent-stack

fastagentstack new myproject --preset standard
uv pip install -r pyproject.toml

The interactive prompt covers database, auth, LLM provider, and deployment options. Use --preset to skip prompts for CI environments.

Run the Dev Server¶

cp .env.example .env
# fill in DATABASE_URL, SECRET_KEY, REDIS_URL

fastagentstack migrate
fastagentstack dev      # auto-reload, binds to 127.0.0.1:8000

Production¶

fastagentstack run      # multi-worker, binds to 0.0.0.0:8000

Or via Docker Compose (generated when include_docker_compose=true):

docker compose up

Adding an Agent¶

# myproject/agents.py
from fast_agent_stack.core.ai.llm.anthropic import AnthropicLLMBackend
from fast_agent_stack.core.ai.llm import Message

_backend = AnthropicLLMBackend(model_id="claude-haiku-4-5-20251001")

def register_agents(app):
    @app.agent("chat", backend=_backend)
    async def chat(messages: list[Message], *, user_id, api_key_id, conversation_id) -> str:
        return f"Echo: {messages[-1].content}"

Then in app.py:

from .agents import register_agents
register_agents(_stack)

CLI Reference¶

Command	Description
`fastagentstack new <name>`	Scaffold a new project
`fastagentstack dev`	Development server (127.0.0.1, reload)
`fastagentstack run`	Production server (0.0.0.0, multi-worker)
`fastagentstack migrate`	Apply framework + user migrations
`fastagentstack makemigrations`	Generate migration from model changes
`fastagentstack worker <module>`	Start Dramatiq worker
`fastagentstack scheduler <module>`	Start Periodiq scheduler
`fastagentstack createsuperuser`	Create a superuser account
`fastagentstack version`	Print version