The Agent Skills Directory

[SAFE]: The skill implements a comprehensive safety-first workflow for landing code changes, requiring mandatory verification of CI status and local test suites before allowing a merge.
[SAFE]: A dedicated Repository Safety Check (Step 3) identifies potential data leakage by scanning PR content for credentials, internal tool references, and infrastructure details before merging into public repositories.
[SAFE]: The instructions explicitly forbid the use of administrative overrides (--admin) to bypass established branch protection rules, ensuring that security policies are respected.
[SAFE]: Post-merge monitoring (Step 6) ensures that any breakage on the target branch is immediately detected and surfaced to the user.
[PROMPT_INJECTION]: The skill creates a surface for indirect prompt injection by ingesting untrusted data from PR bodies and commit messages during the safety scan. This is a low-risk surface given the context of safety verification and the requirement for user confirmation.
Ingestion points: gh pr view output used in the safety scan (SKILL.md).
Boundary markers: Absent for the retrieved text.
Capability inventory: Pull request merging, branch deletion, and shell command execution.
Sanitization: Not explicitly present for the retrieved text.

merge-pr