git-commit
Pass
Audited by Gen Agent Trust Hub on Mar 14, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill uses standard Git commands (
git diff,git log,git commit) required for its intended functionality. These operations are conducted locally within the repository and do not involve remote code execution or unauthorized network access.\n- [PROMPT_INJECTION]: Analysis of indirect prompt injection surface (Category 8):\n - Ingestion points: The skill reads external data from
git diff --stagedandgit log --oneline -10(SKILL.md, Workflow section).\n - Boundary markers: There are no explicit delimiters or instructions to ignore instructions embedded within the git data.\n
- Capability inventory: The skill is authorized to execute
gitcommands via subprocess as documented in the workflow and reference files (references/quick-reference.md).\n - Sanitization: No sanitization or filtering is performed on the repository data before it is processed by the agent. This is considered a low-risk surface inherent to the skill's primary purpose.
Audit Metadata