reverie-cli

The reverie command-line interface.

Commands

reverie run <cmd...>          Run any command with Reverie auto-instrumentation.
reverie status                Ping the configured backend and show its health.
reverie runs list             List recent runs.
reverie runs show <run-id>    Show one run plus its event timeline.
reverie replay <run-id>       Stream a run's events to the terminal (Phase 1 preview).

How reverie run works

reverie run python my_agent.py does three things:

1. Prepends a small bootstrap directory to PYTHONPATH. The directory contains a sitecustomize.py that Python imports automatically before any user code.
2. The bootstrap calls reverie_openai.auto(), which registers a tracing processor with the OpenAI Agents SDK. Your agent code is unchanged.
3. Spawns your command as a subprocess and waits for it to exit, then exits with the same status code.

This works for any Python entrypoint — python script.py, python -m module, python -c "..." — and is the same mechanism opentelemetry-instrument uses.

Limitations of the bootstrap

If your project ships its own sitecustomize.py, our hook tries to import yours afterward but cannot guarantee perfect compatibility. Set REVERIE_NO_SITECUSTOMIZE_CHAIN=1 to disable chaining if it causes issues.

For non-Python commands (reverie run node app.js), the CLI still runs the command but no instrumentation is injected.

Configuration

Install

uv pip install -e ".[dev]"
reverie status