The Agent Skills Directory

[COMMAND_EXECUTION] (MEDIUM): The file 'references/slash-command.template.md' documents a feature where shell commands prefixed with '!' are executed automatically upon loading the command (e.g., !git branch --show-current). This represents a dynamic execution surface.
[PROMPT_INJECTION] (LOW): The skill is vulnerable to indirect prompt injection because placeholders like {{SHELL_COMMAND_1}} are intended to be filled with instructions that are subsequently executed as shell commands. If an attacker influences these inputs, they can achieve unauthorized command execution.
[DATA_EXPOSURE] (LOW): The documented dynamic context mechanism (Category 10) allows for the execution of commands that could access sensitive environment data or system configurations, although no specific malicious targets are hardcoded in the provided files.

command-optimization