Python MCP server that exposes a curated subset of the Rootly incident-management API to any MCP-compatible IDE (Cursor, Windsurf, Claude, etc.). Features dynamic tool generation, pagination defaults, and endpoint whitelisting for security.
https://github.com/Rootly-AI-Labs/Rootly-MCP-serverStop juggling browser tabs during production fires. The Rootly MCP Server brings your entire incident management workflow directly into Cursor, Windsurf, Claude, and any MCP-compatible editor.
When your production system goes down at 3 AM, the last thing you want is to context-switch between your IDE where you're debugging and a web dashboard where you're managing the incident. This MCP server eliminates that friction entirely.
Most development teams lose precious minutes during incidents switching between tools:
Each context switch costs mental overhead when you need to be laser-focused on resolution. This server collapses that workflow into a single interface.
Direct API Access in Your Editor: Query incidents, alerts, team assignments, and action items using natural language right in your IDE. No more Alt+Tab gymnastics between applications.
Smart Context Management: Built-in pagination defaults to 10 items for incident queries, preventing your AI assistant from getting overwhelmed by massive response payloads. The server knows which API endpoints actually matter during incident response.
Security-First Design: Only whitelisted Rootly API endpoints are exposed. You get access to essential incident management functions without opening up administrative or sensitive operations to AI agents.
Zero-Setup Dynamic Tools: The server automatically generates MCP tools from Rootly's OpenAPI specification. New API endpoints become available without manual configuration.
During Active Incidents: Ask your AI assistant "Show me all P0 incidents from the last 24 hours affecting the payments service" and get structured data instantly. Then drill down with "What action items are assigned to my team for incident INC-2024-0142?"
Post-Incident Analysis: Query related alerts and workflow runs to understand cascade failures: "Find all alerts that triggered during incident INC-2024-0095 and show me the associated workflow runs."
Team Coordination: Check team assignments and service ownership without leaving your debugging session: "Which team owns the auth service and who's on-call right now?"
Install via PyPI and add this configuration to your MCP-compatible editor:
{
"mcpServers": {
"rootly": {
"command": "uvx",
"args": ["--from", "rootly-mcp-server", "rootly-mcp-server"],
"env": {
"ROOTLY_API_TOKEN": "<YOUR_ROOTLY_API_TOKEN>"
}
}
}
}
For teams wanting customized endpoint access, clone the repository and modify the allowed_paths variable in server.py. The default whitelist covers essential incident management operations while keeping administrative functions secure.
Built by Rootly AI Labs with Python 3.12+, the server handles pagination intelligently and filters API access appropriately for AI agents. You can run it locally during development or connect to their hosted version for zero-maintenance operation.
The whitelisted endpoints cover everything you need during incident response: incidents, alerts, teams, services, action items, workflows, and user management. Additional paths can be exposed by modifying the server configuration.
When production systems fail, every second counts. This MCP server ensures your incident management workflow moves at the speed of thought, not the speed of browser tabs.