@Pluvio9yte: Claude Code can't read WeChat public account articles. I spent ten minutes solving it and turned it into an open-source skill. Yesterday I asked Claude Code to review a draft of a public account article. I dropped the link in, and WebFetch directly returned "Environment exception, please complete verification." Switched to Jina Re...

Claude Code can’t read WeChat public account articles — so I spent ten minutes fixing it and turned it into an open-source skill.

Yesterday I asked Claude Code to review a draft from a WeChat public account article. I dropped the link into it, and WebFetch directly returned “Environment anomaly, please complete verification.” I tried routing through Jina Reader instead, and got blocked the same way.

Both tools hit the same wall: WeChat detects the request didn’t come from the WeChat app and directly serves a CAPTCHA page — it won’t even give you the HTML.

I looked into WeChat’s blocking logic. The first layer of checking is simple — it looks at the User-Agent in the HTTP request headers for the keyword MicroMessenger.
Normal browser UAs don’t contain that word, so the WeChat server determines you’re an external visitor and returns a prompt page asking you to open it in WeChat.

The fix was straightforward: replace the User-Agent with the WeChat iOS WebView format, include MicroMessenger/8.0.49, and add a Referer: https://mp.weixin.qq.com. Using curl with those headers, it went right through — I got the full 3MB HTML.

Once I had the HTML, I needed to do three things:
Extract the title from the JavaScript variable msg_title, get the author name from the <meta> tag, and grab the body text from the <div id="js_content">.

There’s a gotcha in WeChat’s HTML — msg_title is sometimes wrapped in double quotes and sometimes in single quotes. My first regex only matched double quotes, so the title came back empty. After adding multi-pattern matching, everything extracted correctly.

After writing the script, I ran it against that “井底之硅” article — title, author, publish time, and 7025 characters of body text all came back in clean Markdown format.

The whole script uses only the Python standard library — urllib for requests, re for HTML parsing, and json for output. Zero dependencies, no API key needed, no Playwright or Selenium required — it runs on any machine with Python.

After finishing, I packaged it into two versions.

One is ra-公众号提取 for internal projects — it’s tied to my content creation system’s storage rules. Extracted source files are automatically saved into the .internal/ privacy zone, and downstream workflows for rewriting, topic selection, and video production can directly use them.

The other is the portable version rn-wechat-extract, placed in the rnskill open-source repository. This version isn’t bound to any project structure — the output path is controlled by the --output-dir parameter. It can be installed into any Claude Code / Codex project.

Installation:

npx -y skills add Pluviobyte/rnskill --skill rn-wechat-extract

Or manually copy the skills/rn-wechat-extract/ directory into your project’s .claude/skills/ folder.

Usage is one command:

python3 fetch_wechat.py "https://mp.weixin.qq.com/s/xxxx" --output-dir ./output

It outputs a JSON with the title, author, word count, and file path.

A few edge cases to know:

• If you make many requests from the same IP in a short period, you’ll trigger a slider CAPTCHA — just wait a few minutes and try again.
• Articles that require login or payment can’t be fetched.
• Images use lazy-loaded data-src — the script extracts text only, no image downloading.

This skill solves a very specific problem: when an Agent workflow encounters a WeChat public account link, the user no longer has to manually copy and paste the article content. One command, full-text extraction, ready for downstream skills.

雪踏乌云 (@Pluvio9yte):
Open-source WeChat article parsing skill. Claude Code or Codex cannot directly access WeChat public account links.
So I asked Grok to figure out the principle, and made this skill that lets Agents download and read the content and images of WeChat articles.

Address