jmb-prototype
Pass
Audited by Gen Agent Trust Hub on Nov 19, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest user input and codebase context to generate functional, executable prototypes.
- Ingestion points: As described in
SKILL.mdandLOGIC.md, the agent uses the 'user's prompt' and 'surrounding code' to determine the prototype's logic and structure. - Boundary markers: The instructions do not define delimiters or explicit 'ignore' instructions to prevent the agent from obeying malicious directives embedded in the provided prompts or source files.
- Capability inventory: The generated prototypes are executable artifacts (HTML/JS or framework components) that can interact with the local filesystem (e.g., scratch databases) and are intended to be run via project task runners like pnpm, python, or bun.
- Sanitization: The process lacks specified validation or escaping mechanisms for inputs before they are used to generate the prototype code.
Audit Metadata