Social Corpus Harvest

Social Corpus Harvest is a reusable Codex workflow backed by the open-source yuwanpai2004-create/codex-skills repository. It combines Agent Reach, OpenCLI, the official OpenCLI Chrome extension, and bundled Python orchestration to collect public posts from Xiaohongshu, Douyin, and Weibo. The workflow verifies the target account before collection, strips temporary signed URL parameters, checks authorship where the platform adapter permits it, and writes one Markdown file per post plus a machine-readable source manifest.

Use it when scattered public posts need to become a traceable local archive, knowledge base, RAG source, content-analysis dataset, migration package, or consented AI persona/training candidate set. Do not use it to guess accounts, collect private content, copy comments or fan edits, bypass platform challenges, or obtain passwords, cookies, session tokens, or CAPTCHA answers. Downloading or transcribing audio/video requires a separate user decision. Collection does not itself grant republication, commercial-use, or model-training rights.