nearby-places-search
Pass
Audited by Gen Agent Trust Hub on Oct 1, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADSCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes external data from the Google Places API which could contain adversarial content intended to influence agent behavior.\n
- Ingestion points: The skill accepts
queryandlocationarguments from the user and ingests search results (names, addresses) from the Google Places API inscripts/search_places.py.\n - Boundary markers: Although
SKILL.mdspecifies a presentation format, it lacks explicit boundary markers or instructions to treat the API-provided text as untrusted content.\n - Capability inventory: The skill has capabilities for network communication via
urllib.requestand local file read access to the agent's workspace files.\n - Sanitization: No sanitization or filtering is applied to the place details retrieved from the API before they are passed back to the agent.\n- [EXTERNAL_DOWNLOADS]: The script communicates with
places.googleapis.comto perform geocoding and place searches. This is a well-known service and is documented as the primary function of the skill.\n- [COMMAND_EXECUTION]: The skill invokes a Python script through the Bash tool to execute its search logic, which is standard behavior for its design.
Audit Metadata