commit-conventions
Pass
Audited by Gen Agent Trust Hub on Mar 12, 2026
Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill executes Git CLI commands including
git status,git diff,git add, andgit commit, and is authorized to execute project-specific formatting commands to resolve hook failures. - [PROMPT_INJECTION]: The skill presents a surface for indirect prompt injection by incorporating untrusted data from the repository into its instructions.
- Ingestion points: Reads repository-specific rules from
AGENTS.mdand context fromgit diffoutputs. - Boundary markers: Absent; the instructions explicitly direct the agent to 'apply any git/commit rules' found in the local file without qualification.
- Capability inventory: The skill can stage files, perform commits, and execute arbitrary local formatter commands.
- Sanitization: No sanitization or verification of the content in
AGENTS.mdis performed before the agent treats it as an instruction.
Audit Metadata