The Agent Skills Directory

[COMMAND_EXECUTION]: The skill invokes a local Python script (load_skills.py) to consolidate instruction files based on the user's selected mode (e.g., sql, python). The script logic is securely constrained to a predefined whitelist of internal files and validated mode parameters, preventing directory traversal or unauthorized script execution.\n- [SAFE]: Comprehensive analysis confirms the absence of sensitive data access, external network dependencies, or obfuscated content. The instruction routing and style enforcement (Caveman Mode) are implemented using safe, standard practices within the agent's execution environment. The helper script relies only on built-in Python libraries and local file system access within the skill's own package directory.

bms