The Agent Skills Directory

[COMMAND_EXECUTION]: The skill utilizes git and gh (GitHub CLI) to perform standard repository operations such as branch switching, pulling updates, merging pull requests, and updating issue states.
[DYNAMIC_EXECUTION]: As part of the post-merge verification flow, the skill executes project-specific test or build commands (e.g., native task runners) to ensure the stability of the merged base branch.
[INDIRECT_PROMPT_INJECTION]: The skill parses the pull request body to automatically identify and close linked feature issues.
Ingestion points: Pull request description text retrieved via gh pr view in SKILL.md.
Boundary markers: Absent; the skill relies on searching for specific patterns within the body text.
Capability inventory: Ability to merge pull requests (gh pr merge), close issues (gh issue close), and post comments (gh issue comment).
Sanitization: Extraction is limited to specific keyword patterns (e.g., Closes #<N>), which mitigates broader injection risks.
[SAFE]: The skill contains explicit "Red Line" safety instructions that prevent the agent from force-merging, bypassing branch protections, or proceeding with merges when CI status checks are failing or pending.
[SAFE]: The skill employs secure practices for managing multiline issue comments by writing content to temporary system files (${TMPDIR}) instead of passing large, potentially unescaped strings directly as shell arguments.

merge-pr-verify-close