investigate
Pass
Audited by Gen Agent Trust Hub on Sep 16, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSDATA_EXFILTRATIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill suggests installing 'ripgrep' via Homebrew and the 'google-generativeai' library via pip. These are well-known and trusted tools for searching and AI integration.
- [DATA_EXFILTRATION]: The skill transmits local source code and error messages to the Google Gemini API for analysis. This behavior is the intended primary purpose of the skill and uses a well-known service.
- [INDIRECT_PROMPT_INJECTION]: The skill provides an attack surface for indirect prompt injection by ingesting untrusted source code and including it directly in prompts for an LLM. Evidence: (1) Ingestion points: Source code from local files like 'problematic-file.ts'. (2) Boundary markers: Absent in the prompt templates. (3) Capability inventory: Triggers external LLM analysis. (4) Sanitization: None performed on the ingested code content.
Audit Metadata