reverie-adapter-openai

Zero-config Reverie instrumentation for the OpenAI Agents SDK.

Use

import reverie_openai
reverie_openai.auto()

# ... your normal Agents SDK code below ...
from agents import Agent, Runner
agent = Agent(name="my-agent", instructions="...")
result = await Runner.run(agent, "Do the thing.")

That's it. Every span produced by the SDK is translated into a Reverie CognitiveEvent and posted to the backend (default http://127.0.0.1:8000).

What you get

The full event tree (parentId + depth) is preserved.

Resilience

• Events are posted in batches of up to 50 every ~100ms.
• The emitter runs in a daemon thread with its own asyncio loop — the agent is never blocked.
• If the backend is unreachable, events are dropped silently. The agent always runs to completion.
• A bounded queue (10k events default) protects against memory blow-up.

Configuration

All optional, set as env vars before reverie_openai.auto():

Install

uv pip install -e ".[dev]"
pytest