google-local-services-search
Pass
Audited by Gen Agent Trust Hub on Jul 15, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONDATA_EXFILTRATIONPROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The script fetches data from the vendor's API endpoint at
https://ai-factory.frevana.com. This is the intended behavior of the skill. - [COMMAND_EXECUTION]: The skill uses a bash script that coordinates
curlandpython3. It includes a Python helper script to safely parse and validate arguments, ensuring they are correctly typed (string, bool, etc.) before being encoded into a JSON payload for the API request. - [DATA_EXFILTRATION]: User queries and the
FREVANA_TOKENare sent to the Frevana API. While this involves sending data to an external domain, the domain belongs to the skill's infrastructure and is necessary for operation. - [PROMPT_INJECTION]: The skill ingests external content from the API response, which constitutes an indirect prompt injection surface.
- Ingestion points: The API response is read and outputted to the agent in
scripts/search_google_local_services.sh. - Boundary markers: None are present to delimit the API data from agent instructions.
- Capability inventory: The skill can write files via the
--outputflag and initiate network requests viacurl. - Sanitization: The script validates that the API response is valid JSON before passing it to the agent.
Audit Metadata