yichen-content-archive

Run the bundled douyin_download.py script with a known douyin.com HTTPS link; it uses Playwright to intercept the aweme/detail API response and extracts the direct video URL. Pass --metadata-only to save only metadata without downloading the video file.

Use xiaohongshu_fetch.py with the note URL and an output directory; it anonymously parses window.__INITIAL_STATE__ for structured metadata, then downloads videos, subtitles, or images. Add --skip-media to save only HTML and metadata, and --use-cookie only after explicit per-turn authorization.

No, it strictly refuses keyword search, site crawling, account discovery, and recommendation expansion. It only accepts URLs the user directly provides, URL files, upstream-confirmed candidates, or explicitly specified known collections with defined limits.

No, known X status and Article URLs are read anonymously through FxTwitter first, with Jina Reader as a public fallback. Authenticated fallbacks via OpenCLI or xreach are only listed as options and require explicit per-turn user authorization before execution.

The executors never overwrite existing files; they automatically create a new suffixed directory such as -run-N or a sibling resume directory. The WeChat exporter treats existing directories as read-only checkpoints and writes pending items to a new resume folder.

The local exporter at 127.0.0.1:18901 needs an active session; you must open its dashboard, scan the QR code, and confirm on your phone yourself. The tool never controls the WeChat client or completes the login on your behalf.