git-workflow

Pass

Audited by Gen Agent Trust Hub on Sep 4, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The scripts/branch-status.sh script ingests data from external sources such as Git branch names and GitHub/GitLab pull request titles (via gh or glab CLI). While these are untrusted data points, the script implements a json_escape function to sanitize output when emitting JSON, and the logic is limited to read-only diagnostic reporting of branch states.
  • Ingestion points: scripts/branch-status.sh (reading branch names and forge PR data), scripts/check-commit-msg.sh (validating user-provided commit messages).
  • Boundary markers: Uses TSV and JSON formatting for data separation.
  • Capability inventory: Limited to Git repository information gathering and status reporting.
  • Sanitization: Includes basic JSON string escaping for backslashes and quotes in branch-status.sh.
  • [SAFE]: The skill demonstrates safe practices by deferring secret scanning and history scrubbing to a specialized agent-history-hygiene skill. It also provides instructions for using Git worktrees to isolate parallel agent sessions, preventing file collisions in a secure local environment.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 4, 2026, 06:53 AM
Security Audit — agent-trust-hub — git-workflow