MLflow Skills
Official MLflow agent skills for tracing, evaluating, and debugging LLM applications with agents.
About
MLflow Skills is the official collection of AI agent skills from the MLflow org, focused on GenAI workflows. It covers instrumenting code with MLflow Tracing, analyzing traces and chat sessions, retrieving traces, running agent evaluations with scorers, querying metrics like token usage and latency, and searching the official docs.
The skills install with npx skills add mlflow/skills or a git clone into the agent's skills directory, and are compatible with Claude Code, Codex, OpenCode, and other agentskills.io-format tools. The repo is licensed Apache-2.0.
What it can do
- Instrumenting with MLflow Tracing for Python and TypeScript
- Trace analysis and multi-turn chat session debugging
- Agent evaluation workflows with datasets and scorers
- Metrics queries: token usage, latency, error rates
- Docs search via the llms.txt index
Setup
Run: npx skills add mlflow/skills, or clone the repo into your agent's skills directory.
Prerequisites
An AI coding agent that supports the agentskills.io format, and the mlflow package installed.
Use cases
Works with
Trust score — 8/10
Scored 8/10 (verified tier, metrics-v1). Evidence: Verified via vendor site/docs/GitHub and reputable coverage, wave 86; Trending score 78/100 from researcher assessment.
★ 79 GitHub stars
Sources
Every fact above was checked against these sources. Where a claim comes from the provider and couldn't be independently verified, the text says so.