The Agent Skills Directory

[PROMPT_INJECTION]: Indirect Prompt Injection Surface
Ingestion points: The skill ingests data from the local repository using git status, git diff, and git log in Step 0 and Phase 1, and retrieves metadata via gh pr list in Phase 2.
Boundary markers: Absent. There are no instructions to delimit or ignore instructions that might be embedded in the repository content (e.g., in diffs or commit messages).
Capability inventory: The skill has significant capabilities, including staging files (git add), creating commits (git commit), and pushing to remote repositories (git push).
Sanitization: Absent. Repository content is analyzed and potentially used to generate commit messages without explicit sanitization.
[COMMAND_EXECUTION]: Version Control Operations
The skill executes shell commands using git and gh. It follows security best practices by requiring user confirmation before critical actions and avoiding automated bulk operations like git add -A.

commit-push