Python SDK

For Python agents that aren't behind a stock LLM SDK — custom orchestration, retrieval pipelines, agent frameworks.

If you call Anthropic or OpenAI directly, use the Anthropic or OpenAI patch instead — it's one line.

1. Get an API key

Sign in to Staso and create a key under Settings -> API Keys. Export it:

export STASO_API_KEY=ak_...

2. Install

pip install staso

Python 3.11+.

3. Decorate your agent

import staso as st

st.init(agent_name="support-bot")

@st.agent
def handle_ticket(ticket_id: str) -> str:
    context = fetch_context(ticket_id)
    return draft_reply(context)

@st.tool
def fetch_context(ticket_id: str) -> dict:
    return db.lookup(ticket_id)

@st.agent marks entry points. @st.tool marks tool functions. Both work with async def.

Open the Staso dashboard — every call to handle_ticket becomes a trace.

Multi-turn conversations

Group traces from one chat thread or background job under a single conversation_id:

with st.conversation("user-42-chat-7", user_id="user-42"):
    handle_ticket("t-1")
    handle_ticket("t-2")

Next

• SDK reference — init, decorators, manual spans, conversations, annotations.
• Guard — block bad tool calls.
• HTTP API — non-Python ingest.