webview-bridge-pages

Pass

Audited by Gen Agent Trust Hub on Sep 9, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill documents mechanisms for ingesting untrusted data through URL parameters and bridge messages.
  • Ingestion points: URL query parameters are used as render inputs (page-implementation.md), and the transport adapter in SKILL.md facilitates communication between the web page and the native host.
  • Boundary markers: The instructions recommend using explicit origin checks and schema validation for all inbound messages to prevent unauthorized frames from accessing the bridge.
  • Capability inventory: The skill implements a postToNative function which allows the web content to send data and trigger actions in the native application environment.
  • Sanitization: The skill provides proactive security guidance, recommending that the native side treat all bridge input as untrusted and requiring strict validation of incoming payloads.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 9, 2026, 09:35 PM
Security Audit — agent-trust-hub — webview-bridge-pages