commit-hygiene
Pass
Audited by Gen Agent Trust Hub on Nov 15, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTIONCREDENTIALS_UNSAFE
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted user-supplied data, including Git commit subjects, message bodies, and diffs, to generate its reports. This represents an attack surface where malicious instructions embedded in a commit message could attempt to manipulate the analysis, verdict, or recommendations.
- Ingestion points: Git history evidence (hashes, subjects, messages, and diffs) required in the workflow.
- Boundary markers: None specified for the output segments that include user-supplied text.
- Capability inventory: No active capabilities (subprocesses, network, or file writing) are exposed; the skill only generates advisory text.
- Sanitization: The skill explicitly requires redacting credentials from the output and shell-quoting arguments in suggested backup commands.
- [COMMAND_EXECUTION]: While the skill is instructed not to execute Git commands itself, it generates command strings for the user to run (e.g.,
git branch <unused-ref> <original-tip-sha>and--force-with-lease). It includes security best practices by requiring shell-quoting of variables to prevent command injection at the user's terminal. - [CREDENTIALS_UNSAFE]: The skill includes a specific safety rule to redact credentials found in the supplied history evidence. This is a security-positive feature designed to prevent the accidental exposure of secrets during the audit process.
Audit Metadata