prototype
Pass
Audited by Gen Agent Trust Hub on Aug 22, 2026
Risk Level: SAFECOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill creates an attack surface by ingesting user prompts and project code to generate functional prototypes. 1. Ingestion points: User prompt and existing source code (SKILL.md). 2. Boundary markers: Absent in prompt construction. 3. Capability inventory: Generation of HTML, JS, and React components for local execution. 4. Sanitization: Not specified. The risk is mitigated by the instruction to use isolated modules and shareable single-file formats.
- [DYNAMIC_EXECUTION]: The skill involves creating executable code based on user requirements. The instructions provide security guardrails by recommending in-memory state and stubs instead of real database connections or persistent storage.
- [COMMAND_EXECUTION]: The skill directs the agent to ensure prototypes can be run using project task runners or interpreters like pnpm, bun, and python, which are standard development practices.
Audit Metadata