MCP Server

The Model Context Protocol (MCP) is an open standard for connecting AI assistants to external data sources. Scout’s hosted MCP server gives your agent direct access to your application’s performance data with zero infrastructure on your end.

Server URL: https://scoutapm.com/mcp

Connecting

1. Add the Scout MCP server to your AI tool’s configuration. For example, in Claude Code:

claude mcp add scout https://scoutapm.com/mcp

Other MCP clients like Claude Desktop use a JSON configuration file. Add the following to your config (e.g., claude_desktop_config.json):

{
  "mcpServers": {
    "scout": {
      "url": "https://scoutapm.com/mcp"
    }
  }
}

2. When your agent first connects, you’ll be redirected to Scout’s OAuth authorization page. Log in (if needed) and approve access.

3. Start asking your agent about your app’s performance.

Authentication uses OAuth 2.0, so your credentials are never shared with the MCP client. The server is read-only — your agent can query data but cannot modify anything.

Available Tools

Once connected, your agent has access to the following tools:

Apps & Organizations

• List your applications and organizations

Metrics & Endpoints

• View time-series metrics (response time, throughput, Apdex, error rate, and more)
• List endpoints with performance data
• Get metrics for a specific endpoint

Traces

• Retrieve recent traces for an endpoint
• View full trace details with spans

Background Jobs

• List background jobs with performance data
• Get metrics and traces for specific jobs

Errors

• View error groups and exceptions

Insights

• Surface N+1 queries, memory bloat, and slow query insights