Welcome to Read.Wiki

One documentation source, three ways to read — browser, llms.txt, real MCP.

One-minute AI integration

This site exposes three parallel read paths for AI tools. MCP clients connect to /mcp directly; URL-only tools fall back to llms.txt.

{
"mcpServers": {
  "readwiki": {
    "url": "https://read.wiki/mcp"
  }
}
}

# Lightweight index (page list + MCP access info)
curl https://read.wiki/llms.txt

# Full corpus (pipe straight into a context window)
curl https://read.wiki/llms-full.txt

curl -X POST https://read.wiki/mcp \
-H 'Content-Type: application/json' \
-H 'Accept: application/json' \
-d '{
  "jsonrpc": "2.0",
  "id": 1,
  "method": "tools/list"
}'

What’s inside

Real MCP endpoint

/mcp is a stateless Streamable HTTP JSON-RPC server. It registers every page as a resource, plus the search_docs tool and the answer_with_readwiki prompt.

AI text exports

/llms.txt lists pages, /llms-full.txt concatenates them. Both are generated at build time and readable by any tool that can fetch a URL.

Bilingual + Pagefind search

/zh/ and /en/ are peer trees; ⌘K / Ctrl+K runs Pagefind full-text search with the index built at compile time.

Multi-version deploys

Build /v1.0/, /v2.0/, … from the same tree by setting VERSION. The version switcher in the header derives URLs automatically.

Start exploring

Get Started Run the site locally, tune the config, ship your first page

Core Concepts Three reader paths, how the MCP server works, the llms.txt ecosystem

API Reference Functions, resources, and tools this repo exposes

Tutorials Basics through advanced, end-to-end