anfra-build-data-app
Pass
Audited by Gen Agent Trust Hub on Nov 19, 2026
Risk Level: SAFE
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill facilitates the integration of UI and charting libraries like Basecoat and ECharts from well-known content delivery networks (jsDelivr, Cloudflare). These resources are loaded using pinned versions and mandatory Subresource Integrity (SRI) hashes, ensuring the code remains untampered.
- [DYNAMIC_EXECUTION]: While the skill involves generating HTML and JavaScript, it enforces a security-first approach by requiring the resulting applications to run within a sandboxed frame with restricted permissions (allow-scripts only, opaque origin). This mitigates the risk of unauthorized repository access or malicious behavior at runtime.
- [DATA_EXFILTRATION]: The development guidelines specifically note that the environment's sandbox blocks standard networking APIs (fetch, XHR, WebSockets) to all but a specific set of whitelisted domains, effectively preventing unauthorized data transfer to external servers.
- [INDIRECT_PROMPT_INJECTION]: The skill defines clear boundaries for processing data retrieved from the warehouse via the SDK. It includes instructions for sanitizing data outputs (e.g., using HTML encoding) and suggests using structured data schemas to minimize the risk of malicious instructions being processed as data.
Audit Metadata