返回 MCP 目录
public公开dns本地运行

Rootly监控平台

从您的集成开发环境(IDE)管理事件。一个MCP服务器,允许通过Rootly API提取事件及其相关元数据。

article

README

Rootly MCP Server

An MCP server for Rootly API that you can plug into your favorite MCP-compatible editors like Cursor, Windsurf, and Claude. Resolve production incidents in under a minute without leaving your IDE.

Demo GIF

Prerequisites

  • Python 3.12 or higher
  • uv package manager
    curl -LsSf https://astral.sh/uv/install.sh | sh
    
  • Rootly API token

Run it in your IDE

Install with our PyPi package or by cloning this repo.

To set it up in your favorite MCP-compatible editor (we tested it with Cursor and Windsurf), here is the config :

{
  "mcpServers": {
    "rootly": {
      "command": "uvx",
      "args": ["--from", "rootly-mcp-server", "rootly-mcp-server"],
      "env": {
        "ROOTLY_API_TOKEN": "<YOUR_ROOTLY_API_TOKEN>"
      }
    }
  }
}

If you want to customize allowed_paths to access more Rootly API paths, clone the package and use this config.

{
  "mcpServers": {
    "rootly": {
      "command": "uv",
      "args": [
        "run",
        "--directory",
        "/path/to/rootly-mcp-server",
        "rootly-mcp-server"
      ],
      "env": {
        "ROOTLY_API_TOKEN": "<YOUR_ROOTLY_API_TOKEN>"
      }
    }
  }
}

Features

This server dynamically generates MCP resources based on Rootly's OpenAPI (Swagger) specification:

  • Dynamically generated MCP tools based on Rootly's OpenAPI specification
  • Default pagination (10 items) for incident endpoints to prevent context window overflow
  • Limits the number of API paths exposed to the AI agent

Whitelisted Endpoints

By default, the following Rootly API endpoints are exposed to the AI agent (see allowed_paths in src/rootly_mcp_server/server.py):

/v1/incidents
/v1/incidents/{incident_id}/alerts
/v1/alerts
/v1/alerts/{alert_id}
/v1/severities
/v1/severities/{severity_id}
/v1/teams
/v1/teams/{team_id}
/v1/services
/v1/services/{service_id}
/v1/functionalities
/v1/functionalities/{functionality_id}
/v1/incident_types
/v1/incident_types/{incident_type_id}
/v1/incident_action_items
/v1/incident_action_items/{incident_action_item_id}
/v1/incidents/{incident_id}/action_items
/v1/workflows
/v1/workflows/{workflow_id}
/v1/workflow_runs
/v1/workflow_runs/{workflow_run_id}
/v1/environments
/v1/environments/{environment_id}
/v1/users
/v1/users/{user_id}
/v1/users/me
/v1/status_pages
/v1/status_pages/{status_page_id}

We limited the number of API paths exposed for 2 reasons:

  • Context size: because Rootly's API is very rich in paths, AI agents can get overwhelmed and not perform simple actions properly.
  • Security: if you want to limit the type of information or actions that users can access through the MCP server

If you want to make more paths available, edit the variable allowed_paths in src/rootly_mcp_server/server.py.

About the Rootly AI Labs

This project was developed by the Rootly AI Labs. The AI Labs is building the future of system reliability and operational excellence. We operate as an open-source incubator, sharing ideas, experimenting, and rapidly prototyping. We're committed to ensuring our research benefits the entire community. Rootly AI logo

Developer Setup & Troubleshooting

1. Install dependencies with uv

This project uses uv for fast dependency management. To install all dependencies from your pyproject.toml:

uv pip install .

2. Using a virtual environment

It is recommended to use a virtual environment for development:

uv venv .venv
source .venv/bin/activate

3. Running the test client

To run the test client and verify your setup:

python test_mcp_client.py

5. General tips

  • Always activate your virtual environment before running scripts.
  • If you add new dependencies, use uv pip install <package> to keep your environment up to date.
  • If you encounter issues, check your Python version and ensure it matches the project's requirements.

6. Connecting to Our MCP Server

You can configure your client to connect directly to our hosted MCP server:

{
  "mcpServers": {
    "rootly": {
      "command": "npx",
      "args": [
        "-y",
        "mcp-remote",
        "https://mcp.rootly.com/sse",
        "--header",
        "Authorization:${ROOTLY_AUTH_HEADER}"
      ],
      "env": {
        "ROOTLY_AUTH_HEADER": "Bearer <YOUR_ROOTLY_API_TOKEN>"
      }
    }
  }
}
help

运行方式说明

cloud

托管运行

托管运行通常表示这个 MCP Server 由服务方环境承载,用户一般按页面提供的连接方式或授权流程接入,不需要在本地长期启动一个 MCP 进程

  1. 打开服务方连接页
  2. 完成授权或复制端点
  3. 在 MCP 客户端中连接
terminal

本地运行 / 其它方式

本地运行通常需要用户在自己的电脑或服务器上安装依赖,把 server_config 复制到 MCP 客户端,并按 env_schema 补齐环境变量、密钥或其它配置

  1. 复制 server_config
  2. 安装所需依赖
  3. 补齐环境变量后重启客户端