Overview
What is Rootly MCP Server?
An MCP server that connects AI agents (Cursor, Claude, Windsurf, Gemini CLI, etc.) to the Rootly incident management platform via the Rootly API. It provides tools for managing incidents, alerts, services, on-call schedules, and related operations.
How to use Rootly MCP Server?
Use the hosted version by adding a URL (Streamable HTTP recommended: https://mcp.rootly.com/mcp) to your MCP client configuration. Authentication uses OAuth2 (automated browser login) or a Rootly API token. Optionally run locally with uv by setting ROOTLY_API_TOKEN. Tool availability can be narrowed with query parameters, headers, or environment variables.
Key features of Rootly MCP Server
- Hosted and self-hosted deployment options
- Multiple transports: Streamable HTTP, SSE, Code Mode
- OAuth2 or API token authentication
- Full (200+ tools) and slim (~70 tools) profiles
- Write tools toggle (
ROOTLY_MCP_ENABLE_WRITE_TOOLS) - Workflow‑specific tool subsets (Incident Response, On‑Call, Monitoring)
- Supports Rootly CLI for standalone operations
Use cases of Rootly MCP Server
- Emergency incident response and management from AI chat
- On‑call scheduling and shift coordination
- Monitoring and alerting workflows
- Querying services, teams, escalation policies, and users
- Automating incident‑related tasks via MCP clients
FAQ from Rootly MCP Server
How do I authenticate with the hosted server?
Use OAuth2 (automatic login via browser) or provide a Rootly API token in the Authorization header as a Bearer token.
What transport options are available?
Streamable HTTP (recommended), SSE (stable), and Code Mode (experimental). Each uses a different endpoint path.
Can I restrict which tools are exposed?
Yes. Use the ?tool_profile=slim query parameter, the X-Rootly-Tool-Profile header, or set ROOTLY_MCP_ENABLED_TOOLS to a comma‑separated allowlist of exact tool names.
Do I need to install anything locally?
No. The hosted server works without local installation. For self‑hosting you need Python 3.12+, uv, and a Rootly API token.
What is the difference between full and slim tool profiles?
The default full profile exposes 200+ tools. The slim profile narrows to approximately 70 high‑usage tools, reducing the surface for simpler workflows.
