SwarmTrace

What is SwarmTrace?

Debugging multi-agent LLM systems is painful. When a pipeline fails, you don't know which agent caused it and have to re-run everything from scratch. SwarmTrace fixes this: - Span tree viz — every agent action as an OTel graph - Time-travel replay — fork any step, edit prompt, replay downstream only - LLM-as-judge — auto scoring of each agent's output - WebSocket streaming, OTLP export, PyPI SDK No sign-up — click "Try Demo →" on the login page.