For: ML engineers + data scientists wiring SBO3L into existing Python pipelines

Python — first signed decision in 5 minutes

~5 min · prereqs: Python 3.10+ · Docker (for the daemon)

  1. 1

    Install

    Type-stubs included; works with mypy strict.

    pip install sbo3l-sdk
  2. 2

    Start daemon

    Same Docker image as the Node quickstart.

    docker compose up sbo3l -d
  3. 3

    First decision

    AsyncIO-friendly client; the sync flavour is `Sbo3lClientSync` if asyncio doesn't fit your stack.

    from sbo3l import Sbo3lClient

async def main():
    sbo3l = Sbo3lClient(url="http://localhost:8730")
    decision = await sbo3l.decide(
        agent_id="research-01",
        intent={"kind": "erc20.transfer", "to": "0xabc...", "token": "USDC", "amount": 100},
    )
    print(decision.outcome, decision.receipt.signature[:16] + "...")

import asyncio; asyncio.run(main())
    allow ed25519:9aF3-2bC7-8e...
  4. 4

    Verify offline

    Pure-Python verifier (no FFI); accepts capsule as dict or JSON string.

    from sbo3l.verify import verify_capsule

ok = verify_capsule(decision.capsule)
print("✓" if ok else "✗", "strict-mode")