mcp -weread

This MCP server exposes tools to interact with WeRead data (the Chinese reading platform) and present it to MCP-enabled LLM clients. The core capabilities include retrieving the user’s bookshelf, searching books on the bookshelf, fetching per-book notes and highlights (optionally organized by chapter), and obtaining popular reviews for a book. These tools are designed to integrate with MCP-compatible clients like Claude Desktop, enabling LLMs to reason over your WeRead data in a structured, context-aware way. To use it with Claude Desktop, run the server via npx as shown in the installation guide and configure Claude’s MCP settings to point at the running server. You can choose to authenticate via Cookie Cloud (recommended) or directly via a Weread cookie, and the server will fetch and refresh cookies as needed.

Once set up, you can invoke the tools from your LLM prompts. For example, get_bookshelf returns basic information about all books on your shelf; search_books lets you filter by keyword; get_book_notes_and_highlights provides chapter-organized notes and highlights; and get_book_best_reviews pulls in popular reviews with metadata. These responses are structured to be easy for the LLM to parse and use in downstream reasoning and responses.

Prerequisites:

• Node.js 16.x or higher
• npm (comes with Node.js) or npx
• Access to a Weread account (cookie-based authentication) or a Cookie Cloud account for automatic cookie management

Installation steps:

1. Ensure Node.js is installed:
   • macOS/Linux: curl -fsSL https://install-node.now.sh/lts | bash
   • Windows: install from https://nodejs.org/
2. Install and run the MCP server via npx (no global install required):
   • Open a terminal and run: npm i -g mcp-server-weread (optional if you prefer direct npx usage) npx -y mcp-server-weread
3. Configure the MCP client (e.g., Claude Desktop):
   • In Claude, add a new MCP tool with the following config: { "mcpServers": { "mcp-server-weread": { "command": "npx", "args": ["-y", "mcp-server-weread"], "env": { "CC_URL": "https://cc.chenge.ink", "CC_ID": "您的ID", "CC_PASSWORD": "您的密码" } } } } Note: If you prefer not to use Cookie Cloud, you can provide WEREAD_COOKIE directly in the env block instead of CC_URL/CC_ID/CC_PASSWORD. The server will prioritize CookieCloud when available.

Tips and common issues:

• Prerequisite cookies: Ensure you have a valid WeRead cookie (or configure Cookie Cloud). The server uses these cookies to fetch data from weread.qq.com.
• Cookie refresh: If cookie expires, Cookie Cloud can automatically refresh cookies; otherwise you must update the WEREAD_COOKIE or CC_* env vars.
• Network access: The MCP server needs outbound network access to WeRead and any Cookie Cloud endpoints.
• Environment variables: You can keep sensitive values in a .env file locally or enter them directly in the MCP client configuration; the example above shows inline config, but secret management is recommended for production.
• Compatibility: Works with MCP-enabled LLM clients such as Claude Desktop; ensure the client’s MCP version supports the expected JSON schema and the tool names (get_bookshelf, search_books, get_book_notes_and_highlights, get_book_best_reviews).