prototype
Pass
Audited by Gen Agent Trust Hub on Nov 1, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill defines a process for generating code based on user prompts and project context, creating an attack surface for indirect prompt injection. * Ingestion points: User prompts and project source code (SKILL.md). * Boundary markers: No explicit instructions for delimiters to isolate untrusted data. * Capability inventory: Creation of HTML/JS files and modification of existing UI components and routes (LOGIC.md, UI.md). * Sanitization: No specific sanitization steps are mandated for processing external input.
- [SAFE]: The skill follows security best practices for development workflows. * It mandates that all generated code be clearly marked as 'throwaway'. * It encourages the use of in-memory state or isolated scratch databases/files. * It includes a mandatory cleanup protocol (Rule 6 in SKILL.md) requiring the agent to list all modifications and prompt the user for confirmation, reducing persistence risks.
Audit Metadata