Yes, Arize released the Phoenix MCP Server, which implements the Model Context Protocol (MCP) for Phoenix. It provides a unified interface for managing prompts, datasets, experiments, and projects, and is designed to integrate with IDEs like Cursor and Claude Desktop to assist in generating and visualizing rich spans for your codebase. The server is open-source and can be installed via npm or used directly with npx for IDE integration. For tracing, it supports context propagation and end-to-end trace visualization in Phoenix using OpenTelemetry and OpenInference instrumentation, including full support for distributed tracing across MCP client-server boundaries as of Phoenix 8.26+.
For setup and usage details, see the official documentation and example configurations for IDE integration and tracing instrumentation. Would you like more detail on how to set up or use the MCP server for tracing in your workflow?
Sources:
- <https://github.com/Arize-ai/phoenix/blob/main/js/packages/phoenix-mcp/README.md|Phoenix MCP Server GitHub README>
- <https://arize.com/docs/phoenix/release-notes/04.2025/04.18.2025-tracing-for-mcp-client-server-applications|Release notes: Tracing for MCP Client-Server Applications>
