yanxue

📚 研学方案管理与智能生成技能。支持按城市、学段、景点、主题、时长生成完整的研学课程方案，并提供方案的保存、管理、Word 导出及文件导入导出功能。适用于中小学（1-9年级及高中）的研学旅行课程设计。

This skill is largely coherent with its stated purpose (generate, save, list, export course plans). Before installing or using it: - Audit the included scripts: manage_courses.py and export_word.py are small; verify they match your expectations. In particular, manage_courses.save will read any file you point it to and write files into /home/ubuntu/yanxue_courses — avoid passing sensitive system file paths as the content_path. - Check /home/ubuntu/yanxue_courses permissions and decide whether you want a skill creating persistent files there. Consider changing STORAGE_DIR to a location you control or run the skill in an isolated environment. - Inspect SKILL.md and filenames for hidden/zero-width unicode control characters (the pre-scan flagged these). Remove or normalize suspicious characters to avoid hidden instructions or obfuscation. - Ensure the runtime environment has the Python packages the export script needs (markdown, htmldocx) and prefer installing them from your vetted package sources. - If you plan to enable any automated distribution (the README mentions sending via Feishu), require explicit user consent and verify network endpoints and credentials before allowing transmissions. - If you have low tolerance for on-disk persistence or for code that can read arbitrary file paths, run this skill in a sandbox/container or request a version that prompts before reading external file paths. I am moderately confident in this assessment; additional details that would raise confidence include a provenance/source URL, a signed release, or confirmation that references/ and storage paths are present and limited to the skill's directory.

Type: OpenClaw Skill Name: yanxue Version: 1.0.0 The skill's core functionality for managing and exporting course plans is benign. However, the Python scripts `scripts/export_word.py` and `scripts/manage_courses.py` exhibit significant vulnerabilities. Both scripts directly use user-provided file paths from `sys.argv` without proper sanitization, making them susceptible to path traversal attacks. Specifically, `manage_courses.py` can be prompted to read the content of arbitrary files (e.g., `/etc/passwd`) and save them within the skill's storage directory, and `export_word.py` could be used to write arbitrary `.docx` files to unintended locations. While there is no clear evidence of intentional malicious behavior within the provided code, these vulnerabilities present a high risk for arbitrary file access and potential data exfiltration if exploited by a malicious agent or through prompt injection, especially given the `SKILL.md`'s mention of '文件分发' (file distribution).