git-pr-workflows-git-workflow

Pass

Audited by Gen Agent Trust Hub on Jul 31, 2026

Risk Level: SAFE
Full Analysis
  • [COMMAND_EXECUTION]: The skill uses standard Git and GitHub CLI (gh) commands to perform repository operations, which is consistent with its stated purpose.
  • Evidence: Execution of commands like git status, git diff, git push, gh pr create, and gh pr checks throughout the workflow.
  • Context: All commands target the user's repository ('origin') and are performed via established command-line interfaces.
  • [INDIRECT_PROMPT_INJECTION]: The skill processes repository-local data and instructions, which acts as a potential surface for indirect injection from the codebase being managed.
  • Ingestion points: Reads instructions from AGENTS.md, contribution guidance, and maintainer documentation within the repository (Step 1).
  • Boundary markers: The skill includes instructions to "Confirm every file in scope" and "Inspect the current branch," though it does not specify explicit prompt delimiters for external content.
  • Capability inventory: Access to the shell to run git, gh, and local repository validation scripts.
  • Sanitization: Relies on the user and parallel agent reviews (Step 2) to identify risks in the content and policy instructions.
  • [PROMPT_INJECTION]: No evidence of malicious override patterns or safety bypasses. The instructions contain safety-positive constraints such as "Repository policy wins over flags and user shorthand" and "Do not weaken gates."
Audit Metadata
Risk Level
SAFE
Analyzed
Jul 31, 2026, 04:34 PM
Security Audit — agent-trust-hub — git-pr-workflows-git-workflow