diff-preview
Pass
Audited by Gen Agent Trust Hub on Feb 17, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSPROMPT_INJECTION
Full Analysis
- EXTERNAL_DOWNLOADS (SAFE): The skill installs the 'google-generativeai' library using pip. Since the package is from a trusted organization (Google), the severity is downgraded per [TRUST-SCOPE-RULE].\n- PROMPT_INJECTION (LOW): The skill is susceptible to indirect prompt injection through untrusted data in git diffs. \n
- Ingestion points: Diff output from the
git diffcommand (staged, cached, or between branches) is captured in the$DIFFshell variable.\n - Boundary markers: Absent. The
$DIFFcontent is interpolated directly into the LLM prompt string without delimiters or instructions to ignore embedded commands.\n - Capability inventory: The skill primarily executes git commands and shell scripts but allows the output to be interpreted by an AI agent.\n
- Sanitization: Absent. No escaping or filtering is applied to the git diff content before it is sent to the LLM.
Audit Metadata