skills/madflojo/dotfiles/commit/Gen Agent Trust Hub

commit

Pass

Audited by Gen Agent Trust Hub on Jun 19, 2026

Risk Level: SAFE
Full Analysis
  • [COMMAND_EXECUTION]: Uses standard git commands (git diff, git add, git commit) to manage version control tasks as described in the skill purpose.
  • [INDIRECT_PROMPT_INJECTION]: Ingests untrusted data from the repository via git diff --staged. The skill mitigates potential shell injection attacks by using a non-interactive stdin-based commit flow (`git commit -F
  • <<'EOF'`) instead of passing messages via shell arguments, ensuring that instructions or special characters within the diff are not executed.
  • [DATA_EXPOSURE]: Reads the contents of staged files to generate contextually relevant commit messages. This operation is limited to the local git environment and does not involve exfiltration of data to external services.
Audit Metadata
Risk Level
SAFE
Analyzed
Jun 19, 2026, 08:33 PM
Security Audit — agent-trust-hub — commit