Xhsfenxi Pro

小红书全链路分析 skill。数据采集（SeleniumBase XHR拦截）+ 博主深度分析（三型分类 + 五层模型 + hsword内核三段论）+ 爆款选题公式（6模型 + 30选题方向）+ 结构化报告 + 黑体Word交付。 整合 xhscosmoskill（采集引擎）+ xhsfenxi（分析框架）+ h...

This package appears to be a developer-created Xiaohongshu scraper + analysis tool and largely does what it says, but review before using: 1) It requires SeleniumBase, a compatible browser (Chrome/Chromium) and drivers — those are not declared in the registry; install them in a controlled environment (e.g., a VM or container). 2) The tool needs a login cookie file (xhs_cookies.json) to access authenticated endpoints; cookies are effectively session credentials: treat them like secrets. 3) The README/SKILL.md reference a login script (xhs_login.py) that is not included — confirm how you will obtain safe cookies. 4) The code can post comments (post_comment) as the logged-in user — avoid using your primary account until you trust the code. 5) SKILL.md contains absolute developer paths (Desktop paths, LIB_ROOT) suggesting it's tailored to the author’s machine — change paths to safe locations. 6) The skill writes to files inside its package (data/*.json) and creates Word reports; consider running it in an isolated environment and inspect outputs. If you need to proceed, run in a sandbox, review xhs_login.py or create your own secure cookie generation workflow, and audit third‑party dependencies (seleniumbase and any underlying undetected/UC tooling) before granting credentials.

Type: OpenClaw Skill Name: xhsfenxi-pro Version: 2.0.0 The skill bundle provides extensive Xiaohongshu (XHS) scraping and analysis capabilities using SeleniumBase and XHR interception. It contains high-risk features including an automated commenting function (`post_comment` in `client.py`) which the `SKILL.md` documentation explicitly claims is 'out of scope,' suggesting a discrepancy between stated and actual capabilities. Additionally, the code contains numerous hardcoded absolute file paths to sensitive session cookies (e.g., in `utils.py` and `SKILL.md`), which is a significant security vulnerability. While no clear evidence of data exfiltration or intentional malice was found, the combination of automated interaction capabilities and poor credential management warrants a suspicious classification.