swiftui-webkit
Pass
Audited by Gen Agent Trust Hub on Aug 23, 2026
Risk Level: SAFEPROMPT_INJECTION
Full Analysis
- [PROMPT_INJECTION]: The skill provides instructions for loading and interacting with remote web content, which presents a surface for indirect prompt injection.
- Ingestion points: External data enters the agent's context through
WebView(url:)andWebPage.load()functions defined inSKILL.mdandreferences/wkwebview-fallback.md. - Boundary markers: The documentation explicitly instructs the implementation of
WebPage.NavigationDecidingfor host allowlisting and recommends routing off-domain links to the system browser. - Capability inventory: The skill manages significant capabilities including JavaScript execution (
callJavaScript), script message handlers (userContentController), and custom URL schemes (URLSchemeHandler). - Sanitization: The instructions provide strong sanitization guidance, such as mandating the use of the
argumentsparameter for JavaScript calls to prevent string interpolation/script injection and utilizingWKContentWorldfor execution isolation.
Audit Metadata