us-all/mlflow-mcp-server

Overview

An MCP server that exposes MLflow resources such as experiments, runs, registered models, model versions, logged models (v3), traces, and assessments through the Model Context Protocol. It allows clients to interact with MLflow programmatically over MCP, enabling integration with MCP-compatible agents and tools.

Best for

Best for
Developers building MCP-based automation or agents that need to interact with an existing MLflow tracking server.

Use cases

• Browse and query MLflow experiments and runs via MCP
• Retrieve registered models and model versions through MCP
• Access logged model artifacts and trace data

Notes

An MCP server that exposes MLflow resources such as experiments, runs, registered models, model versions, logged models (v3), traces, and assessments through the Model Context Protocol. It allows clients to interact with MLflow programmatically over MCP, enabling integration with MCP-compatible agents and tools.

0 stars on GitHub. Last updated 2026-06-18. Licensed MIT.

Use cases

• Browse and query MLflow experiments and runs via MCP
• Retrieve registered models and model versions through MCP
• Access logged model artifacts and trace data

Pros

• Covers a comprehensive set of MLflow entities – experiments, runs, models, traces
• Built in TypeScript with clear typing and MCP standards
• Enables tooling interoperability by exposing MLflow via a standard protocol

Cons

• Zero stars and few indicators of community adoption or stability
• Requires a running MLflow server or tracking URI to function
• Limited documentation or examples beyond the repository README

Indexed from awesome-mcp-servers-punkpeye and enriched against its public facts.