The Agent Skills Directory

[PROMPT_INJECTION]: The skill implements input sanitization on user arguments, including length truncation and character stripping. It also uses boundary tags to encapsulate user input when delegating to sub-agents.
[COMMAND_EXECUTION]: Utilizes jq via the Bash tool to manage session state in a local configuration file. Execution is safe as it relies on normalized data derived from sanitized user input.
[SAFE]: No network access, credential exposure, or remote code execution patterns were detected. The skill follows best practices for input handling and local file operations.

start