restaurant-menu-benchmarking
Pass
Audited by Gen Agent Trust Hub on Sep 21, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill utilizes a bundled utility,
scripts/crawlora.sh, to interface with the Crawlora REST API. This script is restricted to performing HTTP GET requests and usescurlfor network communication.\n- [SAFE]: The bundled script implements high-standard security mitigations: Secret Handling (uses a temporary file with restricted permissions to pass the API key tocurl), Route Validation (strict allowlist for API paths), and Input Sanitization (validates API key format and rejects dangerous path characters).\n- [INDIRECT_PROMPT_INJECTION]: The skill processes restaurant menu data from external platforms via the Crawlora API.\n - Ingestion points: API responses from
api.crawlora.netretrieved viascripts/crawlora.sh.\n - Boundary markers: Instructions direct the agent to verify store metadata and inspect specific payload fields.\n
- Capability inventory: Execution of
scripts/crawlora.shwhich performs network operations.\n - Sanitization: Robust path and parameter validation within the shell script helper.
Audit Metadata