Xiaohongshu Operate

小红书运营执行技能。用于按给定素材执行发帖草稿创建（普通图文/长文/视频）、评论区检查与回复流程、关键词搜索并抓取已有内容。适用于用户明确要求“直接按素材操作平台”而不是重新生成内容的场景，默认发帖到草稿箱、评论先汇总后确认、抓取输出为 Top 20 的标准字段 Markdown。

This skill appears to do what it says: automate Xiaohongshu draft creation, comment workflows, and search scraping via UI automation. Before installing, consider: 1) Browser profile access — the skill requires the agent to use a browser profile named 'openclaw' (which will have whatever accounts/cookies are stored). Confirm which profile will be used and that it does not contain unrelated or highly privileged accounts. 2) Screenshots and page snapshots — these may capture account names, private messages or other on-screen sensitive info; ensure you are comfortable with that. 3) File access — only provide image/video paths you intend the agent to upload. 4) Posting behavior — default is to save drafts and require confirmation before sending replies, but verify the confirmation flow in your environment and monitor first runs. 5) Compliance and rate limits — automated actions can trigger platform defenses; use a dedicated automation account or throttle behavior if needed. If you want higher confidence, ask the publisher for details on how the 'openclaw' browser profile is provisioned/isolated, logging/audit behavior, and any runtime network endpoints the agent might talk to (none are declared in SKILL.md).

Type: OpenClaw Skill Name: xiaohongshu-operate Version: 1.0.0 The skill bundle is a legitimate automation tool for Xiaohongshu (Little Red Book) operations, including post drafting, comment management, and content scraping. It contains strong defensive instructions in 'persona.md' and 'references/comment-ops.md' specifically designed to prevent the exfiltration of sensitive data like API keys, .env files, or account credentials, even if a user attempts to solicit them. The browser automation logic is transparent, follows a 'save draft' default for safety, and lacks any indicators of malicious intent or obfuscation.