The Agent Skills Directory

[SAFE]: The skill is designed for internal brand and copy reviews. It operates by reading local project files and returning a summary table. No risky operations such as network access, credential handling, or code execution were identified.
[COMMAND_EXECUTION]: The skill does not contain any shell commands or subprocess execution patterns. It explicitly states it should only report findings and never edit or refactor code.
[INDIRECT_PROMPT_INJECTION]: The skill ingests data from the repository (brand documentation and code diffs). While this presents a surface for indirect prompt injection, the risk is mitigated by the skill's lack of write/execute capabilities and its strict requirement to return a specific output format.

review-brand