Scout Monitoring (local) — Claude Code MCP server for traces
A local MCP server that hands Scout APM data to your coding assistant: app metrics, endpoints, traces, errors, N+1 insights. You get performance fixes suggested right where you work.
claude mcp add scout-mcp-local -e SCOUT_API_KEY=<SCOUT_API_KEY> -- uvx scout-mcp-localWhat Scout Monitoring (local) does
The server pulls Scout Monitoring data through Scout's API and exposes it as MCP tools. Your assistant can read traces and errors with line-of-code details, then target fixes in the editor. Slow endpoints, N+1 queries, memory bloat, throughput problems — the usual suspects show up next to the code that causes them.
It ships as a Docker image, scoutapp/scout-mcp-local, meant to be started by the assistant itself. The tool set includes list_apps, get_app_metrics, get_app_endpoints, get_endpoint_metrics, and get_app_endpoint_traces. Editor-integrated assistants use traces and backtraces to edit code directly. Desktop assistants can build small JS apps to explore whatever data you need.
The server refuses to start without an API key: set SCOUT_API_KEY in the environment or pass it as a command-line argument. It is a read-only key created on Scout's settings page, not the Agent Key. Docker is required. For local trials, the README suggests monitor: true and errors_enabled: true in the agent config so data actually flows.
What it helps with
- Spot N+1 queries from trace data
- Get an error backtrace with the exact code line
- Read response_time and throughput per app
- List all endpoints with aggregated performance numbers
- File a GitHub issue or Jira ticket from an error trace
Example prompts for the agent
- List the Scout APM apps I have access to
- Check
my-appfor N+1 queries in the last 24 hours - Pull metrics for the
/checkoutendpoint and tell me why it lags
How to install Scout Monitoring (local)
Platform CLI
claude mcp add scout-mcp-local -e SCOUT_API_KEY=<SCOUT_API_KEY> -- uvx scout-mcp-localReplace <…> with your own values. uvx runs the command named after the package; if the README names another one, use uvx --from scout-mcp-local <command>
Platform CLI
codex mcp add scout-monitoring-local --env SCOUT_API_KEY=<SCOUT_API_KEY> -- uvx scout-mcp-localReplace <…> with your own values. uvx runs the command named after the package; if the README names another one, use uvx --from scout-mcp-local <command>
Config file
{
"mcpServers": {
"scout-monitoring-local": {
"command": "uvx",
"args": [
"scout-mcp-local"
],
"env": {
"SCOUT_API_KEY": "<SCOUT_API_KEY>"
}
}
}
}Replace <…> with your own values. uvx runs the command named after the package; if the README names another one, use uvx --from scout-mcp-local <command>
Config file
{
"servers": {
"scout-monitoring-local": {
"type": "stdio",
"command": "uvx",
"args": [
"scout-mcp-local"
],
"env": {
"SCOUT_API_KEY": "<SCOUT_API_KEY>"
}
}
}
}Replace <…> with your own values. uvx runs the command named after the package; if the README names another one, use uvx --from scout-mcp-local <command>
Platform CLI
gemini mcp add -e SCOUT_API_KEY=<SCOUT_API_KEY> scout-monitoring-local uvx scout-mcp-localReplace <…> with your own values. uvx runs the command named after the package; if the README names another one, use uvx --from scout-mcp-local <command>
Step by step, with commands for each platform: How to add an MCP server to Claude Code, Codex, Cursor, VS Code, Gemini CLI and more
Questions about Scout Monitoring (local)
How do I install Scout Monitoring (local) in Claude Code?
Run claude mcp add scout-mcp-local -e SCOUT_API_KEY=<SCOUT_API_KEY> -- uvx scout-mcp-local. Replace the placeholder with your actual API key. Another option: claude mcp add --env SCOUT_API_KEY=<SCOUT_API_KEY> --transport stdio scout-monitoring-local -- uvx scout-mcp-local.
What does Scout Monitoring (local) do?
It feeds Scout APM data to your coding assistant: app metrics, endpoints, traces, and errors with line numbers. An editor-integrated assistant can act on that data to fix code; a desktop assistant can build reports or file tickets.
What do I need before running it?
A Scout Monitoring account, an API key from the settings page, and Docker. Without SCOUT_API_KEY the server will not start. The key is read-only and separate from the Agent Key.
