The Agent Skills Directory

[SAFE]: The skill is entirely composed of markdown instructions and metadata. It does not include any scripts, executable code, or remote downloads. The instructions specifically define the agent as a 'read-only specialist' and explicitly forbid editing, staging, committing, or invoking implementation workflows. While the skill processes user-supplied file paths for code review, the restricted 'read-only' scope and lack of write/execution capabilities effectively mitigate risks associated with indirect prompt injection.

snakepolish