wechat-claw-skill

微信公众号文章全自动写作与发布。从信息搜集、AI配图规划、报刊级 HTML 排版、到草稿创建和发布的完整流程。适用于 AI 日报、财经周报、深度分析、新闻资讯类文章。内置多种模板，并提供本地渲染、校验、图片规划和发布流水线脚本。

This skill is internally consistent, but it delegates sensitive actions to external scripts you must supply: the nanobanana image generator and the wechat_mp uploader/publisher. Before using, (1) inspect/review any external scripts (generate_image.py, wechat_mp.py) you pass to run_pipeline — they will be executed with subprocess and could run arbitrary commands or exfiltrate data; (2) verify how the WeChat script stores and uses credentials (prefer env vars or secure stores over plaintext files); (3) be careful when giving the skill URL sources to collect_sources.py — it can fetch remote content you point to; (4) if you need autonomous agent runs, restrict what inputs the agent can provide so it cannot point the pipeline at untrusted scripts or endpoints. Finally, ask the author why 'searxng' is listed as a required skill in metadata since the instructions say the collector will not crawl without explicit sources.

Type: OpenClaw Skill Name: wechat-claw-skill Version: 1.0.0 The skill bundle provides a comprehensive automated pipeline for WeChat article generation and publishing, but it includes several high-risk capabilities. Specifically, `scripts/collect_sources.py` allows for arbitrary local file reading and remote URL fetching (potential SSRF/data exfiltration), while `scripts/run_pipeline.py` utilizes `subprocess.run` to execute external Python scripts based on paths provided as arguments. Although these features are clearly aligned with the stated purpose of a publishing workflow and lack evidence of intentional malice or obfuscation, they constitute a significant attack surface that could be exploited if the agent is manipulated via prompt injection.