wechat-mp-mcp

A Model Context Protocol (MCP) server for crawling WeChat Official Account articles via the mp.weixin.qq.com subscription-account search interface.

Works with any MCP-compatible client (Claude Code, Claude Desktop, Cline, Continue, Cursor, etc.).

What it does

Given a WeChat Official Account name, this server can:

• search for the account and resolve its internal fakeid

• pull the article list (full history on first crawl, incremental afterwards)

• fetch a single article and convert its body to Markdown

• store everything in a local SQLite database (deduped by URL)

Requirements

• Python 3.10+

• A personal WeChat subscription account (订阅号). You log in once via QR code at https://mp.weixin.qq.com/, then this server reuses your login session to call the public-account search interface.

Note: the subscription-account search interface has a daily quota (roughly ~200 calls/day per account). A first-time full crawl of a large public account may need to be spread across multiple days. Incremental crawls cost very few calls.

Install

cd ~/Documents/code/wechat-mp-mcp
python3 -m venv .venv
source .venv/bin/activate
pip install -e .

Step 1 — Log in (one-time, repeat when session expires)

wechat-mp-mcp-login

The script will:

1. open https://mp.weixin.qq.com/ in your browser

2. ask you to paste the dashboard URL after login (it extracts token from it)

3. ask you to paste the Cookie: header value (copy from DevTools → Network)

Credentials are saved to ~/.config/wechat-mp-mcp/auth.json (chmod 600).

Sessions typically last 1–2 hours. When the server returns "session expired", just re-run wechat-mp-mcp-login.

Step 2 — Run the MCP server

wechat-mp-mcp

It speaks MCP over stdio. You won't normally run it by hand — your MCP client launches it.

Step 3 — Wire it into an MCP client

Generic config (works for Claude Desktop, Cline, Continue, etc.):

{
  "mcpServers": {
    "wechat-mp": {
      "command": "/Users/YOUR_USER/Documents/code/wechat-mp-mcp/.venv/bin/wechat-mp-mcp"
    }
  }
}

For Claude Code (CLI):

claude mcp add wechat-mp /Users/YOUR_USER/Documents/code/wechat-mp-mcp/.venv/bin/wechat-mp-mcp

Tools exposed

Typical workflow

1. Find the account: ask the LLM "search WeChat account 'XX'". It calls search_account → you confirm the right fakeid.

2. First-time full crawl: ask "crawl all history". The LLM walks list_articles_page with begin=0, 5, 10, ... until empty. Stop and resume any time — pagination state lives in the DB.

3. Daily incremental: schedule a prompt "pull new posts from 'XX'" — the LLM calls crawl_incremental (cheap, a few API calls).

4. Read a post: "fetch and summarize <mp.weixin.qq.com/s/...>" — the LLM calls fetch_article.

Storage

SQLite at ~/.config/wechat-mp-mcp/wechat.db by default. Override with WECHAT_MP_MCP_DB=/path/to/file.db.

Schema: account(fakeid PK, nickname, ...), article(link PK, fakeid, title, update_time, body_markdown, ...). Inspect with any SQLite client.

Caveats

• The subscription-account search interface is undocumented. The Weixin team can change it at any time; expect to re-tune the request shape occasionally.

• Read counts / likes / "看一看" are not available through this path. Those require intercepting the WeChat App's traffic (out of scope).

• Be polite: keep delay_seconds ≥ 1.0 between requests, or you'll get rate-limited for 24 hours.