Rednote autopilot: AI drafts, Playwright publishes, Chrome logs in
It automates the full Rednote content pipeline—from scraping Weibo trends to scheduled publishing—by driving a browser with your existing Chrome login state.

What it does
xhs_ai_publisher is a desktop and web service that automates content creation and publishing on Xiaohongshu (Rednote). It uses Playwright to drive a real browser for posting, supports AI text generation via OpenAI-compatible or local Ollama endpoints, and can scrape trending topics from Weibo, Baidu, and Bilibili to seed posts. The PyQt desktop app handles visual editing and preview; a FastAPI service mode runs headless for unattended scheduling.
The interesting bit The project treats browser automation as a login-state problem, not just a DOM problem. It can scan your system Chrome profiles to import an existing Xiaohongshu session, avoiding SMS and CAPTCHA re-verification, and it encrypts API keys locally rather than pretending cloud storage is safer. The scheduled publisher even lazy-loads the Playwright runtime until the first request, which is a nice touch for container deployments.
Key highlights
- Dual interface: thick-client PyQt GUI for drafting, plus a FastAPI headless service for API-driven publishing.
- Trend scraping: pulls hot topics from Weibo, Baidu, Toutiao, and Bilibili to generate content ideas.
- Local image generation: produces cover images and multi-page content graphics from built-in marketing templates without calling external image APIs.
- Login-state reuse: imports cookies and
storage_statefrom system Chrome; falls back to manual login if risk controls trigger. - Multi-account support with isolated local data and encrypted key storage under
~/.xhs_system/.
Caveats
- Desktop and Docker paths are finicky: PyQt5 fails on Windows with Python 3.13 or 32-bit interpreters, and Linux browser startup often requires missing system dependencies.
- The README admits that auto-importing Chrome logins will skip if Chrome is already running, and headless Docker deployments cannot handle live CAPTCHAs—meaning a logged-in desktop session must precede any unattended server setup.
Verdict Worth a look if you operate multiple Xiaohongshu accounts or run a content scheduling service. Skip it if you are not targeting the Chinese market or if you consider browser-automation posting to be against platform terms.
Frequently asked
- What is BetaStreetOmnis/xhs_ai_publisher?
- It automates the full Rednote content pipeline—from scraping Weibo trends to scheduled publishing—by driving a browser with your existing Chrome login state.
- Is xhs_ai_publisher open source?
- Yes — BetaStreetOmnis/xhs_ai_publisher is open source, released under the Apache-2.0 license.
- What language is xhs_ai_publisher written in?
- BetaStreetOmnis/xhs_ai_publisher is primarily written in Python.
- How popular is xhs_ai_publisher?
- BetaStreetOmnis/xhs_ai_publisher has 2k stars on GitHub.
- Where can I find xhs_ai_publisher?
- BetaStreetOmnis/xhs_ai_publisher is on GitHub at https://github.com/BetaStreetOmnis/xhs_ai_publisher.