Skip to main content

MCP Server for Developer Documentation

Splitit's MCP server for engaging with developer documentation (i.e., the contents of this site) provides a question-driven, step-by-step flow, which allows you to decide the best way to add Splitit to your or your customer's business.

Endpoint​

https://docusaurus-mcp.sandbox.splitit.com/sse

Connecting to the Server​

Add the JSON below to your Claude Desktop configuration file (claude_desktop_config.json): 

{
  "mcpServers": {
    "splitit-docs": {
      "command": "npx",
      "args": ["mcp-remote", "https://docusaurus-mcp.sandbox.splitit.com/sse"]
    }
  }
}

Tools​

get_onboarding_context

This tool is ALWAYS called first in any conversation with the MCP server. It returns the Splitit integration flow overview plus a required_next_step field that MUST be followed before any other tool is called.

Order of operations:

  1. Presents the primary four Splitit integration method options to the user
  2. Asks which method they want to use
  3. STOPS and wait for an answer
  4. Finally calls get_integration_guide with their chosen method (does NOT call get_integration_guide until the user has explicitly chosen a method)
get_integration_guide

This tool is called immediately after the merchant selects an integration method with get_onboarding_context. It loads the full integration guide for the selected method, which must be one of (case-insensitive):

  • "platform plugins"
  • "checkout solutions"
  • "direct api"
  • "splitit go"
search_docs

This tool performs a semantic search over all Splitit developer docs, including API specs. Use it for questions about integration, payment flows, SDKs, webhooks, or API usage. Note that you should call get_onboarding_context first if you are just beginning your integration conversation. search_docs always returns a title and summary for each result.

fetch_doc

fetch doc allows you to fetch the full content of a document by its ID. It returns id, title, url, type, source, api_endpoint, summary, and the full raw markdown/source content of the document. Use it to read complete documentation including all details, code samples, and notes, i.e., not just a summary as with search_docs. Note that it does NOT include request/response schemas (use fetch_doc_request_schema and fetch_doc_response_schema for those).

fetch_doc_request_schema

fetch_doc_request_schema fetches the request body schema for an API document, returning the schema JSON if available, or an error if it is not found.

fetch_doc_response_schema

fetch_doc_response_schema fetches the response schemas for an API document. status_codes must be a list containing '200' and/or '400' responses only.

Example Chat​

See this Claude chat for an example of when the various tools are called in the context of a integrator conversation.