skills/chenmijiang/ai-notebook/commit/Gen Agent Trust Hub

commit

Pass

Audited by Gen Agent Trust Hub on Mar 10, 2026

Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill executes git diff --cached to read staged code changes and git commit -m to perform the commit. These are standard local operations required for the skill's primary function.
  • [PROMPT_INJECTION]: As the skill ingests data from git diffs (untrusted external input), it has an indirect prompt injection surface. This is mitigated by the 'Confirm and commit' workflow step, which mandates that the user review, edit, or cancel the action before the commit is executed, preventing the agent from performing unintended operations without oversight.
Audit Metadata
Risk Level
SAFE
Analyzed
Mar 10, 2026, 12:46 AM