add-server-logic
Pass
Audited by Gen Agent Trust Hub on Sep 15, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [Indirect Prompt Injection Surface]: The skill includes a feature that automates workspace exploration via a sub-agent to analyze existing source files, configurations, and comments. If these repository files contain unverified or third-party content, there is a possibility that instructions embedded within them could interact with the agent context.
- Ingestion points: Codebase repository files, including
src/**/*.{ts,tsx,js,jsx,vue,astro}and metadata in.powerpages-site/, are read in Phase 1.4. - Boundary markers: Explicit isolation delimiters or specific warnings telling the exploration sub-agent to ignore instructions inside source file strings/comments are absent.
- Capability inventory: The skill utilizes powerful tools including
Write,Edit, andBashwhich are available throughout the lifecycle. - Sanitization: Input content from files is reviewed directly without custom sanitization layers or filtering filters prior to sub-agent ingestion.
Audit Metadata