The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes git and gh (GitHub CLI) commands to retrieve repository status, fetch PR metadata, and generate diffs. Evidence found in Step 1, 2, and 5 of SKILL.md where it runs gh pr view, gh pr diff, and git diff.
[DATA_EXFILTRATION]: The skill reads project context files such as PLAN.md, DECISIONS.md, and QA-REPORT.md and transmits synthesized findings back to GitHub via gh pr comment. While this is intended functionality, it involves sending local data to an external service (GitHub).
[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it processes untrusted data from external PR descriptions and code comments.
Ingestion points: Data enters the agent context via gh pr view and gh pr diff in Step 1, and git diff in Step 2 of SKILL.md.
Boundary markers: Absent. There are no instructions to ignore or treat embedded instructions in the reviewed code as data rather than instructions.
Capability inventory: The skill has the ability to post comments (gh pr comment), modify files (Step 8: 'fix directly'), and commit changes (git commit).
Sanitization: Absent. The skill does not describe any validation or escaping of the content retrieved from Pull Requests before analysis.

review