google-gemini-api
Pass
Audited by Gen Agent Trust Hub on Sep 4, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill contains a utility script
scripts/check-versions.shthat executes standard commands (npm list,grep,sed,node -v) to verify the local environment and SDK versions. These commands are used solely for configuration checking and do not pose a security risk. - [INDIRECT_PROMPT_INJECTION]: The skill provides patterns for processing external multimodal data (images, videos, PDFs) and web search results (via the Grounding feature). This creates an attack surface for indirect prompt injection if the ingested data contains adversarial instructions.
- Ingestion points:
SKILL.mdand reference guides describe tools for reading files and web content. - Boundary markers: The provided templates do not include specific boundary delimiters or safety prompts to isolate external content.
- Capability inventory: The skill utilizes file system reads (
fs.readFileSync) and network communication with Google's Gemini API endpoints. - Sanitization: Sanitization of processed external data is not implemented in the provided examples, relying on the underlying LLM's safety protocols.
Audit Metadata