Feature Request: Support for Model Context Protocol (MCP)

[nirga]

Summary
Add support for the Model Context Protocol (MCP) in openllmetry to enhance compatibility with standardized model context tracking and improve observability for generative AI workflows.

Background
MCP provides a standardized way to define and propagate contextual metadata across AI/ML inference workflows. Integrating MCP with openllmetry can enable better context-aware logging, tracing, and performance analysis.

Proposed Solution
Implement MCP-compatible context propagation in openllmetry, ensuring seamless integration with OpenTelemetry-based tracing.
Map existing openllmetry context attributes to MCP's schema where applicable.
Provide utilities or middleware to facilitate MCP adoption for users of openllmetry.
Ensure compatibility with frameworks like LangChain, LlamaIndex, and other AI model orchestration tools.

Open Questions
What is the best way to structure MCP integration within openllmetry?
Should we provide automatic instrumentation for MCP, or require explicit user adoption?
Are there any existing MCP-compatible libraries we can leverage?

Next Steps
Investigate MCP’s data model and how it aligns with openllmetry.
Design an integration approach (manual vs. automatic context propagation).
Implement a prototype and test with supported AI model frameworks.
Gather feedback from users and iterate.

[gyliu513]

Some reference may help a bit:

• https://github.com/tinybirdco/mcp-tinybird/tree/main/mcp-server-analytics
• https://www.tinybird.co/blog-posts/analyze-mcp-server-usage