The Agent Skills Directory

[COMMAND_EXECUTION]: The skill provides a Python script (scripts/port_skill.py) designed to be executed locally to process and migrate skill bundles. The script's operations are confined to the local filesystem using user-provided paths.
[SAFE]: The implementation follows security best practices for local file manipulation:
Symlink Protection: The reject_links function identifies and blocks symbolic links or non-regular files using lstat, mitigating path traversal and symlink-based attacks.
Resource Budgeting: Enforces strict limits on the number of files (1,000) and total data size (20 MiB) to prevent Denial of Service through resource exhaustion.
Secure File Access: Uses defensive flags like O_NOFOLLOW and O_NONBLOCK when opening files, and employs exclusive write mode (xb) to ensure existing data is never overwritten.
Input Sanitization: Explicitly rejects remote URLs (detected by ://) and validates directory name patterns using regular expressions.
Output Security: Implements a terminal_safe function that escapes control characters in printed output, protecting the user's terminal from potential injection through processed file content.

skill-porter