Cutover's MCP Server

‍

The Cutover MCP (Model Context Protocol) Server is an open-source, MCP-compliant bridge between large language models (LLMs) and Cutover’s operational systems—such as runbooks and tasks.

It enables AI-powered querying and execution of Cutover workflows without building a custom API client, using the standardized MCP format for plug-and-play integration with any MCP-capable AI tool.

‍

Key features

• Natural-language queries for runbooks, tasks, and statuses from MCP-compatible AI clients.
• Operational actions (e.g., start a runbook) directly from your AI environment.
• MCP-standard compliance for universal AI tool compatibility.
• Secure API access with environment-based credentials.

Who is it for? 

• Runbook Admins and Incident Managers who need fast, AI-assisted access to actions.
• Developers building AI-powered workflows.
• Support teams who want to query Cutover without switching tools.
• Anyone who wants to read, interpret, and act on Cutover workflows using AI.

What you can do

• Query runbooks in real time (e.g., “Show me all runbooks tagged as Disaster Recovery”).
• Trigger actions (e.g., “Start the DR-Failover-East runbook”).
  Create, manage and update runbooks (e.g. “Create a runbook template with 5 tasks”)
• Combine with monitoring tools to automatically execute runbooks based on alerts.

How It works

1. A user makes a natural language request to an MCP client such as Github Copilot or Claude Desktop.
2. The LLM driving the MCP client (e.g. Claude Sonnet 4 in Claude Desktop) translates this into a structured MCP request.
3. The MCP Server uses its tools to translate the request and uses the Cutover Developer API to fetch or modify data.
4. The AI tool provides a natural language response back to the end-user.

‍

Why it matters

• AI-Driven Runbook Workflows – AI agents can directly access, interpret, and run operational workflows in real time.
• Faster Insights, Less Toil – Reduce manual effort by letting AI handle common queries and actions.
• Standardized AI Integration – Works with any AI model or tool that supports MCP, ensuring future-proof compatibility.

Example use cases

• Incident response – Pull real-time status of critical runbooks.
• Monitoring automation – Trigger failover workflows from an AI agent based on alerts.
• Reporting – Query recent task completions or failures without UI navigation.

Requirements

• Node.js ≥ 18
• Valid Cutover API key with appropriate permissions
• MCP-compatible AI client

‍

Get started

Ready to start building? Click here for setup instructions, examples, and full documentation.

‍

‍

‍