Agentops MCP Server

@AgentOps-AI

リポジトリ

基本情報

カテゴリ

開発者ツール

トランスポート

stdio

作者

AgentOps-AI

投稿者

koreandinner

Agentops MCP Server について

The AgentOps MCP server provides access to observability and tracing data for debugging complex AI agent runs. This adds crucial context about where the AI agent succeeds or fails.

#observability #tracing #ai #agents

Agentops MCP Server の設定

以下の設定を使って、このサーバーを MCP 対応クライアントに追加してください。

{
  "mcpServers": {
    "agentops-mcp": {
      "command": "npx",
      "args": [
        "agentops-mcp"
      ],
      "env": {
        "AGENTOPS_API_KEY": ""
      }
    }
  }
}

Agentops MCP Server のツール

Authorize using the AGENTOPS_API_KEY. If the API key is not provided and cannot be found in the directory, ask the user for the API key.

Get trace information and metrics by trace_id.

Get span information and metrics by span_id.

Reserved for explicit requests for COMPLETE or ALL data. Get complete trace information and metrics by trace_id.

概要

What is Agentops Mcp?

Agentops Mcp is a Model Context Protocol server that provides access to observability and tracing data for debugging complex AI agent runs. It adds crucial context about where an AI agent succeeds or fails, making it useful for AI developers building and troubleshooting agentic workflows.

How to use Agentops Mcp?

Configure your MCP client (e.g., Claude Desktop) by adding the agentops-mcp server with the command npx agentops-mcp and the environment variable AGENTOPS_API_KEY. You can also install it automatically via Smithery or build it locally from the repository. Use the available tools to authenticate, retrieve traces/spans, and fetch performance metrics.

Features of Agentops Mcp

Authenticate with an AgentOps project API key to get a JWT token
Retrieve project information and configuration settings
Look up trace information by trace ID
Get span details by span ID
Fetch performance metrics for a specific trace or span
Obtain comprehensive trace data including all spans and metrics

Use cases of Agentops Mcp

Debugging why an AI agent fails during a multi-step task
Analyzing performance bottlenecks in agent execution traces
Correlating trace and span metrics to pinpoint errors
Monitoring AI agent runs in production or development environments

よくある質問

What tools does Agentops Mcp provide?

It provides tools for authentication (`auth`), getting project info (`get_project`), retrieving traces and spans (`get_trace`, `get_span`), fetching metrics (`get_trace_metrics`, `get_span_metrics`), and getting complete trace data with all spans (`get_complete_trace`).

What are the requirements to run Agentops Mcp?

Node.js version 18.0.0 or higher and a valid AgentOps API key are required.

How do I install Agentops Mcp?

You can install it via Smithery using the command `npx -y @smithery/cli install @AgentOps-AI/agentops-mcp --client claude`, or build it locally by cloning the repository and running `npm install` and `npm run build`.