The Agent Skills Directory

[PROMPT_INJECTION]: The skill exhibits an attack surface for indirect prompt injection, common in tools that process external source code.
Ingestion points: The skill ingests untrusted data from git diff and pull requests via gh pr diff in SKILL.md.
Boundary markers: There are no explicit delimiters or specific 'ignore embedded instructions' warnings provided in the instructions to separate code diffs from agent instructions.
Capability inventory: The skill can invoke multiple review agents and the /home-fix-review-comments skill, which likely performs file-system write operations to apply code fixes.
Sanitization: No content sanitization or validation is applied to the diff data before it is passed to the reviewer agents.
[COMMAND_EXECUTION]: The skill uses git and gh (GitHub CLI) to retrieve code diffs. These are standard development tools and the usage patterns provided in the instructions follow best practices for retrieving project context.

self-review