WeChat Official Account Markdown Fetcher

Problem

WeChat official account articles are readable in a browser, but the body is usually wrapped in page-specific structures. Saving the raw page can pull in QR-code prompts, authorization dialogs, and sidebar noise. When the content needs to be reviewed, summarized, or edited, it is useful to convert the article body into stable Markdown while preserving links and images.

How it works

This skill targets mp.weixin.qq.com body extraction. It first checks for Python 3.10+, then uses scrapling_fetch.py to fetch the page and tries preferred containers such as #js_content and .rich_media_content. If those miss, it falls back to generic selectors like article and main, converts the HTML with html2text, keeps links and images, trims common WeChat footer noise, and truncates output by max_chars for side-by-side review.

When publishing to Feishu, it calls md_to_feishu_doc.py with the generated .json path so the title is read from structured output instead of being improvised by the model. The script connects to Feishu OpenAPI under the current agent identity, can grant edit permission via user_open_id, and can update an existing document with doc_id. Outside Feishu, it replies directly with the Markdown body, title, and source link.

Notes

It requires a usable Python environment and network access to the target URL. If the page depends heavily on JavaScript and the body is empty, a dynamic fetcher variant is needed. Feishu writing does not re-fetch missing covers; cover issues should be fixed upstream.