incident-io

This MCP server integrates Incident.io data with the Model Context Protocol, exposing endpoints to read and create incidents within Incident.io via MCP tooling. The server requires an API key to access Incident.io and will listen on the configured port. You can run it directly with Node using the local build output, or deploy it via your preferred method using the same configuration. Once running, you can query endpoints to health-check, retrieve recent incidents, fetch severities, or create new incidents through the MCP interface.

Prerequisites:

• Node.js (v12+ recommended)
• npm (comes with Node.js)
• Access to an Incident.io API key

Install from GitHub (local development):

1. Clone the repository git clone https://github.com/stuartphilp/incident-io-mcp-server.git
2. Install dependencies cd incident-io-mcp-server npm install
3. Build the project npm run build

Run locally (from built output):

1. Start the server node ./dist/index.js

Alternatively, run from a pre-built distribution or using a package-based approach after installation:

1. Install globally (as in the README) npm install -g github:stuartphilp/incident-io-mcp-server
2. Run the server (global install) incident-io-mcp-server

Environment variables needed:

• API_KEY: Your Incident.io API key Optional:
• PORT: Port to listen on (default 3000)

Example configuration for MCP (local installation): { "mcpServers": { "incident-io": { "command": "node", "args": ["./dist/index.js"], "env": { "API_KEY": "your-api-key-here", "PORT": "3000" } } } }

Notes and tips:

• Ensure API_KEY is kept secret and not committed to version control.
• If using a custom port, ensure it’s not in use by another service.
• When running locally, you can rebuild with npm run build after code changes.
• The endpoints exposed include health checks and MCP paths to list, create, and manage incidents via Incident.io integration.
• If you encounter authentication errors, verify the API key has the necessary permissions in Incident.io.
• For containerized deployments, substitute the Node run command with a suitable docker or uvx/Python-based deployment as needed.