| name | wechat-collect |
| description | Fetch a public WeChat article URL, archive the raw HTML, and convert the article into a stage-1 compatible brief in `content-production/inbox/`. Use when Codex needs to collect公众号文章素材 or start the Stage 2 collect-to-create pipeline from a public `mp.weixin.qq.com` URL. |
WeChat Collect
Collect a public WeChat article and transform it into a brief that can be passed directly to case-writer-hybrid.
Quick Start
Run the default command:
.venv/bin/python -m skill_runtime.cli run-skill wechat-collect --input content-production/inbox/20260403-wechat-collect-url.txt
Prepare Input
Pass a text file containing at least one URL. The first detected URL is used.
Example input file:
content-production/inbox/20260403-wechat-collect-url.txt
Follow Collection Workflow
- Fetch the public article HTML from the first detected URL.
- Extract title, author, date, and candidate正文段落 from the page.
- Build a stage-1 compatible brief that downstream writing steps can reuse.
- Archive the raw HTML for traceability and later extraction tuning.
Write Output
Write the brief to:
content-production/inbox/<date>-<slug>-gzh-brief.md
Write the raw archive to:
content-production/inbox/raw/wechat/<date>-<slug>.html
Respect Constraints
- Only works for publicly reachable article URLs
- Deleted articles or anti-crawl variants may produce reduced-quality extraction or fail explicitly
- Current extraction is usable for pipeline intake, but still needs quality tuning for cleaner argument mining
Read Related Files
- Shared runtime:
skills/wechat-collect/runtime.py
- Pipeline entry:
skill_runtime/engine.py
- Stage 2 workflow:
workflows/stage2-wechat-pipeline.json
- Planning reference:
docs/content-skills-implementation-plan.md
