Open-source Codex skill scrapes Xiaohongshu related keywords, scoring 93 terms per seed word
vista8 · x · 2026-09-30
Developer yaojingang open-sourced two Codex skills for scraping autocomplete/related-search keywords from Xiaohongshu and Baidu (repo: yao-geo-skills).
How it works
- Install the skill in Codex and input one or more seed words;
- Codex drives a browser to search each word, capture dropdown suggestions, then recursively query first-layer terms to get a second layer, saving a real screenshot, list position, and query path for every term;
- After dedup, the AI scores each keyword 1–10 for commercial intent by industry. A test with the seed "留学公司" yielded 110 paths and 93 unique keywords — agency/comparison terms scored high, recruiting/filing terms scored low;
- A buildworkbook.py script exports an Excel workbook with five sheets: keywords, capture log, screenshot index, deduped keyword library, and to-capture queue.
In the author's test, Codex scraped 10,000 Baidu related-search long-tail keywords in about an hour. Useful for free SEO/GEO keyword mining.
More from coding & agent
- Codegraph rockets up GitHub Trending with 72k stars: pre-indexed code knowledge graph for AI coding agents — colbymchenry · 2026-09-30
- MCP official servers repo nears 91k stars on GitHub as ecosystem keeps growing — modelcontextprotocol · 2026-09-30
- Ex-Googler: TensorFlow is in maintenance mode — here's how to migrate to PyTorch — lmoroney · 2026-09-30
- LLM reverse-engineering of SNES games is real but slow, dev explains why — gandamu_ml · 2026-09-30
- Building Lumen in Rust: adding data-carrying enums and match without touching the backend — Abhishekcur · 2026-09-30
- Pushing Claude-generated Remotion videos beyond text: real product UI demos — EagleApprehensive · 2026-09-30