skills/denysdovhan/agents/commit/Gen Agent Trust Hub

commit

Pass

Audited by Gen Agent Trust Hub on Jul 9, 2026

Risk Level: SAFE
Full Analysis
  • [SAFE]: The skill utilizes standard version control commands such as git log, git diff, and git commit. These operations are performed within the local repository context and are essential for the skill's primary purpose. No dangerous command patterns or unauthorized network activity were detected.- [PROMPT_INJECTION]: The skill instructs the agent to analyze local files like README.md and CONTRIBUTING.md to identify commit style conventions. This creates an indirect prompt injection surface.
  • Ingestion points: Local repository documentation and commit logs.
  • Boundary markers: Not explicitly defined in the instructions.
  • Capability inventory: git add, git commit.
  • Sanitization: No explicit sanitization or filtering is applied to the content read from the repository files. This is documented as a potential attack surface, but is considered a low-risk, expected behavior for this type of development tool.
Audit Metadata
Risk Level
SAFE
Analyzed
Jul 9, 2026, 01:48 PM
Security Audit — agent-trust-hub — commit