prototype

Pass

Audited by Gen Agent Trust Hub on Jul 27, 2026

Risk Level: SAFECOMMAND_EXECUTIONDATA_EXFILTRATIONPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill directs the agent to create and inject execution scripts into project task runners such as package.json, Makefile, justfile, and pyproject.toml. These scripts facilitate running the generated prototypes via pnpm, python, or bun commands.
  • [DATA_EXFILTRATION]: The agent is instructed to analyze existing project files and surrounding code to understand logic and state requirements. This involves reading project context from the local filesystem.
  • [PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it incorporates logic derived from potentially untrusted project code and user prompts.
  • Ingestion points: Surrounding project files and user instructions.
  • Boundary markers: None; the skill lacks mechanisms to distinguish between code context and instructions.
  • Capability inventory: Modification of project configuration files and creation of new local scripts.
  • Sanitization: No validation or sanitization of source code is required before the agent processes it for logic generation.
Audit Metadata
Risk Level
SAFE
Analyzed
Jul 27, 2026, 05:58 AM
Security Audit — agent-trust-hub — prototype