agent-pwa-specialist
Pass
Audited by Gen Agent Trust Hub on Oct 2, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes user-supplied data without sufficient isolation or sanitization protocols.
- Ingestion points: The
$ARGUMENTSplaceholder at the end ofSKILL.mdallows raw external data to be interpolated directly into the agent's instruction context. - Boundary markers: No delimiters (such as XML tags or unique string markers) are used to separate the skill's instructions from the user-provided data.
- Capability inventory: The skill's defined capabilities include potential access to sensitive operations such as file system writes (for service worker and manifest generation), network operations (for push notification services), and command-line execution (for Workbox CLI and build tools).
- Sanitization: The skill contains no instructions or logic for escaping, filtering, or validating the content provided via the input placeholder.
Audit Metadata