MLflow Prompt Registry

MLflow Prompt Registry is an MCP server that connects Claude Desktop and other Model Context Protocol clients directly to an MLflow Prompt Registry instance. Machine learning engineers and LLM application developers use this server to centralize prompt engineering workflows, bringing versioned templates from MLflow straight into interactive AI environments. By implementing the MCP Prompts specification, the server enables users to list available prompts across registries and retrieve compiled templates populated with specific variable arguments. This eliminates manual copy-pasting of prompt definitions and ensures that conversational agents utilize officially tracked, team-standardized prompts. The server interacts with an active MLflow server over HTTP, defined through standard tracking configuration, making it practical for teams standardizing prompt evaluation, deployment, and testing across repetitive conversational workflows.

Category: AI & LLM Tooling

Tags: machine learning, mlflow, prompt-engineering, prompt-management

Visit MLflow Prompt Registry

How to install and configure MLflow Prompt Registry

1. Ensure an MLflow server is running with the Prompt Registry available: bash pip install mlflow>=2.21.1 mlflow server --port 5000 2. Clone the repository and build the MCP server: bash npm install npm run build 3. Configure Claude Desktop by editing claude_desktop_config.json: json { "mcpServers": { "mlflow": { "command": "node", "args": ["<absolute-path-to-this-repository>/dist/index.js"], "env": { "MLFLOW_TRACKING_URI": "http://localhost:5000" } } } } 4. Update MLFLOW_TRACKING_URI with your actual MLflow server address and restart Claude Desktop.

What you can do with MLflow Prompt Registry

• Loading version-controlled prompt templates directly into Claude Desktop for repetitive development tasks. - Compiling MLflow prompts with custom JSON argument variables before executing client-side workflows. - Discovering and browsing team-managed prompt catalogs using cursor-based pagination and search filters. - Standardizing production prompts across engineering teams by reading directly from a central MLflow tracking server.

Key facts

• Open Source
• https://github.com/B-Step62/mcp-server-mlflow
• AI & LLM Tooling, Developer Tools & Code Intelligence
• machine learning, mlflow, prompt-engineering, prompt-management

Part of MCP Servers

What is MLflow Prompt Registry?

MLflow Prompt Registry MCP server connects MCP-compatible clients to an MLflow server. It implements the standard MCP Prompts specification, allowing users to discover prompt templates stored in MLflow and compile them directly into AI chat sessions.

What tools are provided by this server?

The server provides two primary tools: list-prompts to search and paginate through registered prompts, and get-prompt to fetch and compile a specific template using variable inputs.

Which MCP clients work with MLflow Prompt Registry?

The server works with Claude Desktop and any client that supports standard Model Context Protocol prompt specifications. Configuration requires setting the Node.js execution path and tracking environment variables.

Is MLflow Prompt Registry open source?

Yes, MLflow Prompt Registry is an open-source project hosted on GitHub at https://github.com/B-Step62/mcp-server-mlflow, allowing developers to inspect, modify, and build the source code locally.