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 --staged and git 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 git commands 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
Risk Level
SAFE
Analyzed
Mar 14, 2026, 09:35 PM