The Agent Skills Directory

[DATA_EXPOSURE]: The skill is designed to read local skill-related files (SKILL.md, scripts, and reference documents) to evaluate their quality and structure. This is the intended core functionality and does not target sensitive user credentials or system configurations.
[COMMAND_EXECUTION]: The workflow allows for modifying the local file system to implement optimization plans. However, the skill strictly enforces a manual review step where the agent must present a plan and wait for the user to explicitly confirm modification commands before proceeding.
[PROMPT_INJECTION]: By processing instructions and documentation from other skills (untrusted data), the skill has an indirect prompt injection surface. This risk is effectively mitigated by the 'Reviewer' design pattern and the mandatory human-in-the-loop confirmation before any proposed changes are applied.

skill-optimizer