Security audit
xhs-html-creator
Security checks for vulnerabilities and agentic risk
Overview
The skill mostly matches its stated purpose (generate Xiaohongshu HTML and screenshots) but contains several implementation inconsistencies (hardcoded user/workspace paths, a hardcoded Feishu user id and cross-skill call, undeclared runtime dependencies) that could lead to unintended data exposure or surprise behavior.
This skill appears to implement the stated Xiaohongshu HTML + screenshot workflow, but review before installing: - Inspect send_images.py and the referenced feishu_image.py script (skills/feishu-image/scripts/feishu_image.py). Confirm where images would be uploaded and whether any credentials are used or stored there. A hardcoded Feishu user id is present — remove or parameterize it. - Remove or replace hardcoded workspace path (C:\Users\95116\.openclaw\workspace). Hardcoded absolute paths can leak files or fail on other machines. - Confirm Playwright and browser binaries are acceptable in your environment. The scripts use sync_playwright but the skill does not declare dependencies; ensure you install dependencies in an isolated environment and understand browser downloads. - Verify xhs-material-collector behaviour and confirm it does not pull data from external sites if you expect only local素材库 usage. - Run the skill in an isolated/test workspace first; monitor network traffic and the subprocess calls to ensure no unexpected data exfiltration to external endpoints. - Consider requiring explicit environment variables or user prompts for any external delivery (Feishu tokens, recipient id) rather than accepting hardcoded values. If you can't or won't audit these points, treat this skill as untrusted — the inconsistencies could be sloppy engineering or could lead to unintended disclosure of files or images.
SkillSpector findings are pending for this release.