AgentOps API MCP Server logo

AgentOps API MCP Server

by AgentOps-AI

The AgentOps API MCP Server allows LLM agents to interact with the AgentOps API for accessing trace data. It provides tools for agents to list recent traces and get detailed information about specific traces.

View on GitHub

Last updated: N/A

What is AgentOps API MCP Server?

This MCP server acts as a bridge between LLM agents and the AgentOps API. It handles authentication and provides tools for agents to access and analyze trace data from their AgentOps projects.

How to use AgentOps API MCP Server?

The server can be used with Claude Coder by configuring the config.yaml file with the server's path. Alternatively, it can be run standalone using the provided run-server script or directly with Python. Agents interact with the server through tool calls, providing their AgentOps API key for authentication.

Key features of AgentOps API MCP Server

  • Secure two-step authentication using JWT tokens

  • Provides tools for listing recent traces

  • Provides tools for retrieving detailed trace information

  • Supports custom API URLs for non-production environments

Use cases of AgentOps API MCP Server

  • Debugging and understanding LLM agent behavior

  • Analyzing the performance of LLM agent tasks

  • Identifying bottlenecks in LLM agent workflows

  • Monitoring the health and stability of LLM agents

FAQ from AgentOps API MCP Server

How does the server handle authentication?

The server uses a two-step authentication process: the agent provides their API key, which the server exchanges for a JWT token from the AgentOps API.

How do I configure the server for use with Claude Coder?

Add the server's path to the mcp_servers section of your Claude Coder configuration file (~/.config/claude-cli/config.yaml).

Can I use a custom AgentOps API URL?

Yes, you can override the default API URL by providing the AGENTOPS_API_URL parameter in your tool calls.

What kind of information can I get about a trace?

You can get detailed information about the trace, including all spans, metadata, and timing information.

Is there a Docker image available?

Yes, you can build and run the Docker container using the provided docker build and docker run commands.