gemini-batch
Pass
Audited by Gen Agent Trust Hub on Oct 14, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes user-provided JSONL files to generate prompts for the Gemini API, creating a surface for indirect prompt injection where malicious instructions within input files could influence model behavior.
- Ingestion points: Data is read from user-specified JSONL files in
scripts/create_batch.jsandscripts/get_results.js. - Boundary markers: The skill does not implement specific delimiters or 'ignore' instructions to isolate the ingested data from the prompt structure.
- Capability inventory: The skill has capabilities to read local files, write local files, and perform network operations via the
@google/genailibrary. - Sanitization: No explicit sanitization, filtering, or validation of the input prompt content is visible in the provided scripts.
Audit Metadata