google-search
Pass
Audited by Gen Agent Trust Hub on Sep 29, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill ingests and processes search results from an external API, creating a surface for indirect prompt injection where adversarial content in search snippets could attempt to influence the agent's logic.\n
- Ingestion points: Untrusted search result data is retrieved from the Frevana API in scripts/search_google.sh.\n
- Boundary markers: The skill instructions do not specify the use of delimiters or clear boundaries when the agent processes the retrieved search results.\n
- Capability inventory: The skill possesses network access (curl), local file system access (writing JSON results), and shell execution capabilities (bash, python3).\n
- Sanitization: The script validates the JSON structure of the API response but does not perform sanitization or filtering of the text content within the search results.
Audit Metadata