Resources

LLMs.txt / AI docs

Machine-readable documentation for AI coding tools and agents. Following the llmstxt.org convention, we publish the docs as plain-text files an LLM can ingest in full — no HTML to scrape.

Directory overview

Each file is served from the site root. Give a tool the whole file, or the scoped one that matches your task, so it answers from the current docs instead of guessing.

File	Contains
llms.txt	The index: a link map of every doc page and the files below.
llms-full.txt	The complete documentation (product, REST API, SDKs, MCP, guides) in one file.
llms-api.txt	The REST API reference only: endpoints, parameters, output formats, errors.
llms-sdks.txt	Language-integration snippets: Node, Python, PHP, Go, Ruby, and OpenAPI codegen.
llms-mcp.txt	The MCP server only: install, tools, and Claude Desktop config.

Which file?

Start with llms-full.txt for the whole picture. Pick a scoped file (llms-api.txt, llms-sdks.txt, llms-mcp.txt) when you want to keep the context small and focused.

Using the files

Point any AI tool at a file’s URL, or paste its contents into the context. A few common setups:

Cursor

Add the URL under Settings → Features → Docs (the @Docs feature), then reference it in chat with @Docs. Cursor indexes the file and pulls the relevant parts into context.

Windsurf

Reference a file with @ in the chat, or add its URL to your .windsurfrules so every session has it.

Claude, ChatGPT & Copilot

Paste the file contents into the conversation, or attach the URL where the tool supports fetching. For the Claude API or Claude Code, fetch the file and include it as context.

fetch the full docs

curl https://serp.hoangha.shop/llms-full.txt

Prefer tools over pasted docs

For an agent that will actually run searches (not just answer questions about the API), connect the MCP server instead — it exposes the API as callable tools rather than reference text.