OpenHands is an open-source AI coding agent framework that autonomously edits files, runs terminal commands, and browses the web. Because these agents operate without human oversight, observability and governance are critical. MLflow integrates with OpenHands via OpenTelemetry to capture structured traces of every LLM call, tool invocation, and token spend. Beyond tracing, MLflow's evaluation toolkit provides 60+ built-in LLM judges (e.g., RelevanceToQuery, Correctness, ToolCallEfficiency) to assess output quality either through the UI or programmatically. Finally, MLflow AI Gateway acts as a centralized proxy for all LLM traffic, enabling budget controls, usage tracking, secret management, and fallback routing with minimal configuration changes.

6m read timeFrom mlflow.org
Post cover image
Table of contents
What is OpenHands? ​Trace OpenHands Agents via OpenTelemetry ​Evaluate OpenHands Agent Runs ​Governance for LLM Traffic with AI Gateway ​Summary ​

Sort: