Skip to content
MCP Server Directory logo
developer-tools

Plesk MCP

MCP server for managing Plesk hosting servers through AI assistants.

Reviewed overview

MCP サーバー概要

what is Plesk MCP? Plesk MCP is a Model Context Protocol (MCP) server that lets AI assistants and apps manage Plesk hosting servers. It integrates with the Plesk REST API (and WP Toolkit) to provide core administration operations, shell command execution, and file uploads while running locally so credentials and sensitive data remain in your environment.

how to use Plesk MCP?

  • Install the uv Python package manager and run the MCP server (examples use uvx/uv commands).
  • Configure an MCP client (e.g., VS Code mcp.json) to start the server, providing PLESK_HOST and credentials (PLESK_API_KEY recommended, or PLESK_USERNAME/PLESK_PASSWORD).
  • Optionally add --insecure when connecting to non-TLS test instances (not recommended for production).
  • For development: uv run plesk-mcp --log-level debug --insecure. See uvx plesk-mcp@latest --help for runtime options.
  • Testing: use the official Plesk Docker image to run a local Plesk instance, or run the MCP in demo mode (mocked responses). Automated tests use uv run pytest, with smoke tests selected via -m smoke.

key features

  • Manage Plesk servers through AI agents that use the Plesk REST API and WP Toolkit API
  • Execute shell commands on target servers and upload files
  • Runs locally to keep sensitive data in your environment
  • Support for multiple Plesk servers concurrently
  • Demo mode with mocked responses for development without a Plesk instance
  • Integration recommendation: pair with a documentation MCP (local or remote) for more accurate agent behavior
  • Standard development tooling: ruff, mypy, pytest, uv build/publish, mcp-publisher

use cases

  1. Remote administration of single or multiple Plesk servers via AI assistants
  2. Automating common hosting tasks (site creation, SSL, backups, user management)
  3. Troubleshooting servers by running shell commands through an agent
  4. CI/CD or operational workflows that need scripted Plesk interactions
  5. Development and testing with Docker-based Plesk instances or demo mode

FAQ

  • What Plesk versions/OS are supported?

Any currently supported Plesk and OS (Linux/Windows) are expected to work; tested on Plesk Obsidian 18.0.76. Ensure API access is enabled.

  • How should I provide credentials?

API key is recommended (create with plesk bin secret_key --create). You can also use username/password. Use env files to avoid embedding credentials in configs.

  • Is it secure?

The MCP server runs locally so data stays in your environment. Avoid using --insecure in production; prefer API keys and TLS.

  • Can I test without a real Plesk instance?

Yes: use the official Plesk Docker image or run the MCP in demo mode (mocked responses).

  • Where is the project hosted / how to get support?

Source and issues are on GitHub: https://github.com/FlayedOne/plesk-mcp