The Agent Skills Directory

[COMMAND_EXECUTION]: The skill manages career profiles by executing local Python scripts (scripts/render_report.py and scripts/validate_report.py) to transform structured JSON data into HTML reports. It also interacts with the lark-cli for document and whiteboard creation in the Lark (Feishu) platform.
[SAFE]: The report generation script (render_report.py) implements a robust FragmentSanitizer using a strict whitelist of HTML tags and attributes. This prevents the injection of executable code or malicious scripts into the generated career profiles.
[SAFE]: Generated HTML reports are configured with a highly restrictive Content Security Policy (CSP) meta tag (default-src 'none'), which prevents the reports from initiating any network connections or loading external resources, ensuring offline safety.
[SAFE]: The skill instructions strictly mandate privacy protection, requiring the agent to remind users to hide sensitive information (ID numbers, addresses, company secrets) and to de-identify data before performing any web searches.
[SAFE]: Persistent data is stored locally in the 职业档案/ directory, which the skill explicitly instructs to be excluded from version control via .gitignore, minimizing the risk of accidental data exposure.

career-planning