Overview
AgentOps MCP Server provides access to observability and tracing data for debugging complex AI agent runs, adding crucial context about where agents succeed or fail. It exposes a dedicated MCP server (agentops-mcp) that can be configured in your MCP config to run via Node.js tooling (npx). The server offers tools for authenticating requests, retrieving traces and spans, and compiling complete trace views with metrics. With 'auth', you can authorize using an AgentOps project API key to obtain a JWT token; 'get_trace' and 'get_span' let you fetch trace and span details by ID; and 'get_complete_trace' returns comprehensive trace information including all spans and their metrics. This enables developers and operators to instrument AI workflows, monitor performance, and quickly diagnose issues in agent behavior. Requirements include Node.js >= 18.0.0 and supplying an AgentOps API key for tool usage. Installation and usage sections describe how to install via Cursor, Smithery, or local development, making it accessible for both staging and production environments.
Features
auth
Authorize using an AgentOps project API key and return JWT token.
get_trace
Retrieve trace information by ID.
get_span
Get span information by ID.
get_complete_trace
Get comprehensive trace information including all spans and their metrics.
Who Is This For?
- Developers:Use MCP to access observability data and debug AI agent runs.
- Observability Engineers:Integrate tracing data into monitoring workflows for AI services and debugging.
- AI Researchers:Study traces and spans to understand agent behavior in experiments.
