Frequently Asked Questions
Everything you need to know about importing specs, generating MCP servers, and connecting them to your AI clients.
MCP Studio turns an OpenAPI or Swagger specification into a runnable Model Context Protocol server. Upload a JSON spec and it parses every endpoint, validates schemas, and generates server code your AI agents can call as tools.
MCP is an open protocol from Anthropic that standardizes how AI applications expose tools, data, and context to models. Generated servers implement protocol revision 2024-11-05 and work with any MCP-compatible client.
OpenAPI 3.0, 3.1, and Swagger 2.0 in JSON format. Only .json files are accepted.
Servers can be generated in TypeScript, Python, Java, and PHP. Pick a runtime and target API base URL before generating.
No. Upload the spec, choose a language, and download the bundle. You only need the matching runtime installed to execute the generated server.
Each bundle includes a ready-to-paste config snippet. Add the server command to your Claude Desktop config or Cursor MCP settings, then restart the client so it discovers the new tools.
No. Base URLs and secrets are supplied through environment variables, so nothing sensitive is hard-coded into the generated server.
Specifications are stored in your local database and addressed by UUID. You can revisit them any time from the API Specs library.
Yes. The Generated Servers catalog tracks every build, so you can re-download or regenerate without uploading the spec again.
The runtime for your chosen language (Node.js, Python, a Java JDK, or PHP) plus network access to the target API base URL.
The most common causes are invalid JSON, a missing top-level paths object, or an unsupported file type. Correct the spec and upload it again.
Uploads are bounded by your server's PHP upload_max_filesize and post_max_size settings.
Still have questions?
Read the official Model Context Protocol documentation or start generating with your own specification.