Rapid7 InsightOps MCP Server

An MCP (Model Context Protocol) server that wraps the Rapid7 InsightOps REST API for log search and analysis. Designed for use with Devin Web (app.devin.ai) as an STDIO MCP server, enabling AI agents to search, query, and analyze production logs directly.

Region: EU only (eu.rest.logs.insight.rapid7.com). Other regions are not supported.

Installation

# From PyPI (once published)
pip install rapid7-mcp-server

# Or directly via uvx
uvx rapid7-mcp-server

# Or from source
pip install git+https://github.com/YOUR_ORG/rapid7-mcp-server.git

Devin Web Setup

1. Open app.devin.ai and go to Settings > MCP Marketplace
2. Click "Add Your Own"
3. Paste this configuration:

{
  "transport": "STDIO",
  "command": "uvx",
  "args": ["rapid7-mcp-server"],
  "env_variables": {
    "RAPID7_API_KEY": "<your Rapid7 API key>"
  }
}

4. Save. The server will start automatically when Devin uses any Rapid7 tool.

Environment Variables

If RAPID7_API_KEY is not set, the server starts but every tool returns a clear error message.

Available Tools

LEQL Quick Reference

LEQL (Log Entry Query Language) is used in the query parameter:

# Text search (case-insensitive substring)
where(error)
where(NullPointerException)

# Key-value search
where(traceId=abc-123-def)
where(statusCode>=400 AND statusCode<500)

# Regex search
where(/timeout.*connection/i)

# Boolean operators
where(error AND NOT healthcheck)
where(level=ERROR OR level=FATAL)

# Statistical queries (returns counts/aggregates, not log lines)
calculate(COUNT)
where(level=ERROR) calculate(COUNT)
where(statusCode>=500) calculate(COUNT) groupby(service)
calculate(AVERAGE:responseTime) groupby(endpoint)

Time Range Examples

Relative: last 5 minutes, last 1 hour, last 24 hours, last 7 days, yesterday, today

Absolute: Use from_timestamp and to_timestamp (UNIX milliseconds).

Development

# Install with dev dependencies
pip install -e ".[dev]"

# Run tests
pytest tests/

# Run the server locally (for testing)
RAPID7_API_KEY=your-key rapid7-mcp-server

License

MIT