google-ai-overview
Pass
Audited by Gen Agent Trust Hub on Jun 26, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill connects to ai-factory.frevana.com to fetch AI Overview data. This is the vendor's official API endpoint and is consistent with the skill's primary purpose. No sensitive data beyond the required authentication token is transmitted to this endpoint.
- [COMMAND_EXECUTION]: The script uses curl and python3 for making network requests and performing data validation. These commands are implemented with proper argument handling and use temporary files for data staging to minimize security risks.
- [PROMPT_INJECTION]: The skill processes data from a remote AI service, which constitutes an indirect prompt injection surface. However, the script is designed for data retrieval and does not execute the content of the response.
- Ingestion points: The API response from the Frevana endpoint is captured into a temporary file.
- Boundary markers: None; the skill returns the raw JSON response to the agent context.
- Capability inventory: Subprocess execution (curl, python3) and file system writes (via the --output flag).
- Sanitization: The script includes a Python-based validation step to ensure the returned data is valid JSON before it is presented to the user or agent.
- [SAFE]: Credentials management for the FREVANA_TOKEN is handled via environment variables or interactive prompts in terminal sessions, following standard security practices for API-based tools.
Audit Metadata