GetMCP turns an existing API into a Model Context Protocol (MCP) server so AI clients can use its tools without requiring the team to write MCP protocol code. It can build tools from cURL commands, OpenAPI or Swagger files, and Postman collections, then lets you review and test them before connecting a server to clients such as Claude, ChatGPT, VS Code, or Cursor. The product is self-hosted, with a WordPress plugin and a standalone app for sites that do not use WordPress.
Key features
API imports and tool creation
GetMCP parses API definitions and generates tools with schemas, descriptions, and parameter validation. The page also describes a manual editor for defining tools with JSON Schema, attaching an HTTP request as the backend, and adding response transformations. This gives teams a route for endpoints they would rather configure by hand or not expose directly.
The import process is intended to avoid building and maintaining the MCP layer yourself. If an API has no OpenAPI or Swagger specification, the page says a cURL command or API documentation can be used to create tools. GetMCP imports API definitions, not application code or a GitHub repository.
Testing, monitoring, and access
A built-in playground allows tools to be tested before they are exposed to clients. In production, the dashboard records calls with details such as tool name, client, latency, status, and payload size. Calls can be filtered, replayed, or exported to JSON or CSV, which can help with troubleshooting and reviewing usage.
The product supports per-client rate limits and configurable analytics retention. Authentication credentials are passed through to the underlying API rather than stored by GetMCP, according to the product page. It describes support for API keys and OAuth 2.1, with the existing API’s permission boundaries remaining in effect.
WordPress and standalone options
The WordPress plugin is designed for installation on a WordPress site or subdomain. The standalone application is self-hosted and does not depend on WordPress; its stated server requirement is PHP 8.3 or newer. SQLite is available without additional database configuration, and MySQL and PostgreSQL are also supported. Servers can be moved between the plugin and standalone app using an export file, according to the page.
Who it is for
GetMCP is suited to SaaS teams, agencies, consultants, WordPress or WooCommerce operators, and internal IT or operations groups that already have APIs they want to make available through MCP clients. It may also be useful to developers and no-code builders who need a way to expose selected endpoints without implementing protocol handling themselves. Agencies can use separate servers for customers or environments, while internal teams can connect private APIs to a self-hosted copilot.
The product is less relevant to someone looking for a general-purpose AI assistant or a tool that creates an application from scratch. Its central role is to connect API-backed tools to MCP-compatible clients.
Limitations and considerations
GetMCP depends on having an API and enough information to define its endpoints and authentication. It does not import or deploy a code repository. Teams without an existing API, usable documentation, or cURL examples may need to prepare those first. The standalone route also has a specific PHP version requirement, while the WordPress option requires a WordPress site. Because the product is self-hosted, users should be prepared to operate it on their own infrastructure.
Verdict
GetMCP focuses on a clear technical task: reducing the work involved in making existing API endpoints available to MCP clients. Its import options, generated schemas, testing playground, and call logs cover the path from initial setup through ongoing troubleshooting. The choice between a WordPress plugin and a standalone app broadens where it can run, while self-hosting keeps API traffic within the user’s environment. Its value depends on having APIs worth connecting and being comfortable maintaining the host where GetMCP runs.
