skills/dydoc/skills/prototype/Gen Agent Trust Hub

prototype

Pass

Audited by Gen Agent Trust Hub on Sep 12, 2026

Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTIONPERSISTENCEDYNAMIC_EXECUTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill instructions in SKILL.md and UI.md direct the agent to run prototypes using local task runners including 'pnpm', 'python', and 'bun'.
  • [INDIRECT_PROMPT_INJECTION]: The skill processes user-supplied requirements to generate executable scripts. 1. Ingestion points: User prompts and surrounding code context are used to determine prototype shape as described in SKILL.md. 2. Boundary markers: There are no instructions to use delimiters or ignore-previous-instruction markers to isolate user requirements from the generated logic. 3. Capability inventory: The agent writes files to the local directory and executes shell commands via project runners. 4. Sanitization: No validation or sanitization procedures are defined for the input data used in code generation.
  • [PERSISTENCE]: SKILL.md suggests that if a prototype requires persistence, the agent should create a local file with a name indicating it is a temporary prototype file.
  • [DYNAMIC_EXECUTION]: LOGIC.md and UI.md specify the generation of JavaScript and TypeScript modules that are executed at runtime in a browser or local environment.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 12, 2026, 06:28 AM
Security Audit — agent-trust-hub — prototype