prototype-designer
Pass
Audited by Gen Agent Trust Hub on Oct 15, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTIONDYNAMIC_EXECUTION
Full Analysis
- [COMMAND_EXECUTION]: The skill executes shell commands based on the scripts defined in the repository's package.json file to validate the generated prototype.
- Evidence: Phase 3.1 instructs the agent to run 'lint', 'dev', 'start', and 'serve' scripts found in package.json.
- Evidence: Phase 3.1 utilizes package manager executors like 'npx', 'pnpm exec', or 'bunx' to run 'tsc --noEmit' for type checking.
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted PRD and User Journey documents to drive the logic and content of the generated UI code, creating a vulnerability to indirect injection.
- Ingestion points: Reads external PRD and User Journey markdown files in Phase 1.1.
- Boundary markers: No explicit boundary markers or 'ignore embedded instructions' warnings are defined for processing these inputs.
- Capability inventory: The skill has the ability to write to the file system (Phase 2.1) and execute shell commands (Phase 3.1).
- Sanitization: No evidence of input sanitization or escaping was found in the instructions.
- [DYNAMIC_EXECUTION]: The skill dynamically generates React, Vue, or Angular source code and proceeds to execute it in a development environment.
- Evidence: Phase 2.2 describes the implementation of pages and components based on external inputs.
- Evidence: Phase 3.1 involves starting the development server to run the newly generated code for verification.
Audit Metadata