Xiaohongshu Ops

小红书端到端运营：账号定位、选题研究、内容生产、发布执行、数据复盘。 Use when: (1) 用户要写小红书笔记/帖子, (2) 用户说"发小红书"/"写个种草文"/"出一篇小红书", (3) 用户讨论小红书选题/热点/爆款分析/竞品对标, (4) 用户提到账号定位/人设/内容方向规划, (5) 用户要求生成...

What to check before installing / using this skill: - Confirm where account credentials live: the SKILL.md assumes the runtime has an authenticated browser profile and access to image-generation APIs (Gemini/idealab, Seedream). Ask the publisher which tokens/keys are required, how they should be provided, and request minimal-scope credentials. Do not provide high-privilege or long-lived keys until you understand use. - Verify the runtime tools and scripts: the instructions call <WORKSPACE>/scripts/generate-image.sh and other scripts and reference using --ref flow. Those scripts are not included in the package. Ask where they come from (platform-provided, installed separately, or part of another repo). If you must add scripts, review them first. - Test in a sandbox account: because the skill automates posting and replies, run it against a test Xiaohongshu account (and a test Feishu workspace, if used) to confirm behavior and avoid accidental public posts or replies. - Confirm data flows and storage: the skill writes prompts, anchors, and knowledge-base files. Decide if those files may contain sensitive data and where they are stored. If you have compliance constraints, audit what gets persisted and who can read it. - Restrict autonomous actions initially: since autonomous invocation is allowed by default, consider disabling or requiring user confirmation for publish/reply actions until you trust the skill and have tested its behavior. - Ask the author for a minimal dependency list: request an explicit list of env vars, API endpoints, and required binaries. The absence of declared credentials for image services and messaging endpoints is the main incoherence here. - Validate source and maintainers: the registry metadata owner and README references point to third-party repos; confirm the publisher's identity and look for an official homepage or repo you can review. If you cannot verify the author, be cautious. If you want, I can produce a short checklist/email you can send to the skill author asking for (1) required env vars and their scopes, (2) the scripts referenced and their contents, (3) where account logins are expected to come from, and (4) a description of what the skill will write to the workspace.

Type: OpenClaw Skill Name: xhs-ops Version: 1.0.1 The xhs-ops bundle is a comprehensive toolkit for Xiaohongshu (XHS) account management, including content generation, account auditing, and automated interaction. It is classified as suspicious because it contains functional JavaScript snippets for scraping user and post data (references/xhs-eval-patterns.md) and explicitly directs the agent to execute external Python and shell scripts (style-observe.py, generate-image.sh) for style tracking and image generation that are not provided in the bundle. While these features align with the stated purpose of the skill, the combination of scraping capabilities and external script execution without providing the source code for those scripts poses a potential risk for unauthorized data collection or local code execution.