caveman-commit

Pass

Audited by Gen Agent Trust Hub on Sep 5, 2026

Risk Level: SAFENO_CODE
Full Analysis
  • [SAFE]: The skill consists entirely of markdown-based instructions for the AI agent. No scripts, configuration for tools, or binary files are present in the skill package.
  • [INDIRECT_PROMPT_INJECTION]: Evaluation of ingestion of untrusted git diffs:
  • Ingestion points: Git diff data (SKILL.md).
  • Boundary markers: Absent.
  • Capability inventory: None; the skill explicitly forbids the execution of git commands (SKILL.md).
  • Sanitization: Absent. The lack of capabilities prevents potential exploits.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 5, 2026, 01:59 AM
Security Audit — agent-trust-hub — caveman-commit