张雪峰AI助手

提供张雪峰风格的高考志愿、专业选择、院校分析及职业规划咨询，回答直白实在、有态度有观点。

This skill is coherent with its stated purpose (local KB + style prompt) but has a few red flags you should consider before installing: 1) SKILL.md/README direct you to download an encrypted knowledge package from vdoob.com — verify the source and authenticity before downloading or running anything. 2) The skill will decrypt and extract that archive into ~/.zhang-xuefeng/knowledge_base (note README and code disagree about an exact path); extraction uses zipfile.extractall without path sanitization, so a malicious or tampered archive could overwrite files or write outside the intended directory. 3) The code includes pycryptodome (used) and 'requests' (unused) in dependencies — unnecessary packages increase surface area. 4) If you proceed, inspect the downloaded package (and its file list) in a safe environment or sandbox before activating; prefer running the skill in an isolated VM/container and avoid using passwords that are reused elsewhere. 5) If possible, ask the publisher for an official, signed release or checksum for the KB and clarification on the expected download path and activation flow.

Type: OpenClaw Skill Name: zhang-xuefeng-ai Version: 1.0.1 The skill implements a gated content model that directs users to an external website (vdoob.com) to download an encrypted 'knowledge base' and purchase activation codes. The extraction logic in `src/kb_manager.py` uses `zipfile.extractall()` without path sanitization, which is a known ZipSlip vulnerability that could allow a malicious ZIP file to overwrite arbitrary files on the user's system. While the code lacks explicit malicious payloads, the combination of a third-party binary requirement and vulnerable extraction logic poses a significant security risk.