git-commit

Pass

Audited by Gen Agent Trust Hub on Nov 19, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill processes repository diffs and status to generate commit messages. This is an inherent attack surface for indirect prompt injection if the code being committed contains adversarial instructions.
  • Ingestion points: Local file diffs and git status (SKILL.md).
  • Boundary markers: Absent.
  • Capability inventory: Executes standard git commands like add and commit (SKILL.md).
  • Sanitization: Absent.
  • [SAFE]: The skill instructions focus on standard Git workflows and explicitly discourage dangerous operations like force pushing or bypassing pre-commit hooks, ensuring that Git hooks are allowed to execute normally.
Audit Metadata
Risk Level
SAFE
Analyzed
Nov 19, 2026, 01:13 AM