conventional-commits

Pass

Audited by Gen Agent Trust Hub on Sep 2, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
  • [SAFE]: The skill instructions are focused on formatting Git-related strings (commits, branches, PRs). No obfuscation, data exfiltration, or unauthorized command execution patterns were found.- [INDIRECT_PROMPT_INJECTION]: The skill possesses an attack surface where untrusted data (git diffs) could influence agent output. \n
  • Ingestion points: git status and git diff as mentioned in the 'Making a commit' section (SKILL.md). \n
  • Boundary markers: The skill includes a 'Verify' section with a checklist that requires the agent to re-read the drafted message against the spec and the actual diff before committing. \n
  • Capability inventory: The agent is instructed to use git commit, git checkout, gh pr create, and gh issue create. \n
  • Sanitization: No explicit sanitization or escaping of the diff content is described before it is used to draft the message.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 2, 2026, 11:54 AM
Security Audit — agent-trust-hub — conventional-commits