restaurant-food-delivery-research
Pass
Audited by Gen Agent Trust Hub on Oct 20, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: Fetches data from the official Crawlora API at
api.crawlora.net. This is a vendor-owned resource used to retrieve restaurant, delivery, and grocery information. - [COMMAND_EXECUTION]: Utilizes a bash script (
scripts/crawlora.sh) to perform network requests. The script implements several security hardening measures: - Strict validation of the
CRAWLORA_API_KEYenvironment variable format to prevent command injection. - A hardcoded base URL to prevent redirection of API keys to malicious hosts.
- Comprehensive whitelisting of allowed API routes and HTTP methods.
- Use of temporary curl configuration files to avoid exposing secrets in the process list.
- Explicit prevention of local file disclosure by blocking the
@character in query arguments. - [INDIRECT_PROMPT_INJECTION]: Processes external content such as user-generated restaurant reviews and menus from platforms like Yelp and OpenTable.
- Ingestion points: Data retrieved from
api.crawlora.netviascripts/crawlora.sh. - Boundary markers: The skill documentation indicates the API returns clean JSON for agent processing.
- Capability inventory: The helper script is restricted to network requests and stdout output; it does not perform file writes or execute arbitrary commands based on input.
- Sanitization: The script sanitizes its environment and enforces a strict route whitelist to ensure only intended endpoints are accessed.
Audit Metadata