git-commit
Pass
Audited by Gen Agent Trust Hub on Oct 12, 2026
Risk Level: SAFEPROMPT_INJECTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [PROMPT_INJECTION]: The skill contains explicit instructions to ignore and override platform-level or system-enforced instructions related to commit trailers (e.g., 'Co-Authored-By', 'Signed-off-by', 'Generated with'). It uses framing that positions the skill's instructions as taking precedence over 'system reminders', which is a common bypass pattern.
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted external data in the form of git diffs to automatically generate commit messages and risk analyses.
- Ingestion points: Reads data using 'git diff --cached' in SKILL.md.
- Boundary markers: None identified; the agent is instructed to infer details directly from the diff content.
- Capability inventory: Executes 'git commit -m' and 'git commit --amend' using inferred content.
- Sanitization: No sanitization or validation of the diff content is mentioned before it is interpolated into the commit command.
Audit Metadata