StuMason/coolify-mcp

What it solves

This project provides a Model Context Protocol (MCP) server that allows users to manage their self-hosted Coolify PaaS (Platform-as-a-Service) using natural language through AI clients like Claude, Cursor, or any other MCP-compatible tool. It eliminates the need to manually navigate the Coolify dashboard for common deployment, debugging, and operational tasks.

How it works

The server acts as a bridge between an AI assistant and the Coolify v4 API. It exposes 44 consolidated tools that the AI can call to perform actions such as deploying applications, managing databases, updating environment variables, and checking system health. To optimize for LLM context windows, the server uses consolidated action-parameter tools and returns summarized projections of resources rather than raw API responses.

Who it’s for

DevOps engineers, developers, and system administrators who use Coolify to host their own applications and want to integrate their infrastructure management into their AI-powered coding and chat environments.

Highlights

• Comprehensive Toolset: Includes tools for infrastructure overview, diagnostics, batch operations, server/project/environment/application/application logs, database management, and cloud provider integration (Hetzner/DigitalOcean).
• Token-Optimized: Reduces tool definitions to ~6,600 tokens (down from ~43,000) to save context window space.
• Safety Guardrails: Implements human-in-the-loop confirmations (elicitation) for destructive operations like deleting resources or stopping all apps, and masks secrets at the API boundary by default.
• Verified Deploys: The deploy tool can poll for terminal status and return log tails on failure, providing more reliable confirmation of success than simple HTTP responses.
• Documentation Search: Integrates a full-text search across Coolify's own documentation to help the AI assistant provide better troubleshooting advice.