commit-work
Pass
Audited by Gen Agent Trust Hub on Sep 14, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONDYNAMIC_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data from the repository's working tree through commands like
git diffandgit diff --cached(documented inSKILL.mdworkflow steps 1 and 4). While this is necessary for code review, it creates a surface where malicious instructions embedded in the code being committed could influence agent behavior.\n - Ingestion points: Output of
git diff,git status, andgit diff --cachedinSKILL.md.\n - Boundary markers: No explicit delimiters or warnings are provided to the agent to treat the diff content strictly as data.\n
- Capability inventory: The skill executes shell commands (git) and potentially arbitrary local build/test scripts (step 7 in
SKILL.md).\n - Sanitization: No sanitization of the diff output is performed before processing.\n- [DYNAMIC_EXECUTION]: The workflow includes a step to 'Run the smallest relevant verification' such as unit tests, lint, or build commands (
SKILL.mdworkflow step 7). This involves executing scripts and code found within the local repository, which is a standard developer practice but represents execution of local code.
Audit Metadata