prototype
Pass
Audited by Gen Agent Trust Hub on Sep 4, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes user prompts and codebase context to generate functional prototypes, which could allow adversarial instructions in the input to be reflected in the resulting code.
- Ingestion points: The agent utilizes user prompts and surrounding code context to determine prototype logic and UI structure (SKILL.md).
- Boundary markers: Absent; the instructions do not require the generated code to use delimiters or ignore embedded instructions within user-provided data.
- Capability inventory: The skill directs the agent to create HTML files, JavaScript modules, UI components, write to the filesystem, and perform git operations (SKILL.md, LOGIC.md, UI.md).
- Sanitization: Absent; no specific sanitization, escaping, or validation steps are prescribed for data ingested from the user during the code generation process.
- [COMMAND_EXECUTION]: The skill explicitly instructs the agent to provide and use shell commands for running prototypes (e.g.,
pnpm,python,bun) and managing version control (committing to branches), which involves executing code and interacting with the system environment.
Audit Metadata