google-related-questions
Pass
Audited by Gen Agent Trust Hub on Oct 2, 2026
Risk Level: SAFECOMMAND_EXECUTIONDYNAMIC_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill executes a local shell script
scripts/search_google_related_questions.shto perform API requests. The script usescurlto communicate with the Frevana API atai-factory.frevana.com. Authentication credentials are provided via environment variables or explicit flags, ensuring that secrets are not hardcoded or exposed in command history. - [DYNAMIC_EXECUTION]: The shell script
scripts/search_google_related_questions.shuses Python heredocs for processing data. This is a secure method to ensure that JSON construction and validation are handled by a dedicated parser rather than fragile shell string interpolation, which effectively mitigates injection risks. - [INDIRECT_PROMPT_INJECTION]: The skill ingests data from an external API, which is a potential surface for indirect injection.
- Ingestion points: The external API response is ingested by the
scripts/search_google_related_questions.shscript. - Boundary markers: The
SKILL.mdinstructions explicitly define the expected JSON response shape and instruct the agent to provide validated results or summaries. - Capability inventory: The script
scripts/search_google_related_questions.shincludes functionality to write the API response to a local file path provided by the user. - Sanitization: The script includes a validation step that uses Python to verify the response is valid JSON before it is processed or saved, reducing the risk of malformed data exploitation.
Audit Metadata