short-term-rental-market-research
Pass
Audited by Gen Agent Trust Hub on Oct 20, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill uses a bundled helper script (
scripts/crawlora.sh) to perform API requests. The script is highly defensive, featuring strict format validation for the API key, a regular-expression-based allow-list for API routes, and path-traversal prevention. It handles the API key securely by using a temporary configuration file with restricted permissions (umask 077), which is automatically cleaned up on exit. - [INDIRECT_PROMPT_INJECTION]: The skill ingests external data from the Crawlora API, which serves as a potential surface for indirect prompt injection.
- Ingestion points: Market aggregate and listing data are retrieved from
api.crawlora.netas described inSKILL.mdandreference/endpoints.md. - Boundary markers: None explicitly defined in the provided instruction templates.
- Capability inventory: The skill can execute the local bash helper script to perform network operations via
curl. - Sanitization: The helper script enforces route allow-listing and uses
curl --data-urlencodeto escape query parameters, effectively mitigating basic injection attempts during the data retrieval phase.
Audit Metadata