webg-spec-to-goal
Pass
Audited by Gen Agent Trust Hub on Jul 14, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill follows secure software development practices for scaffolding tools. It utilizes well-known, legitimate packages such as Phaser, Vite, and Playwright, and provides a transparent, well-documented project structure.
- [PROMPT_INJECTION]: The skill ingests untrusted user input in the form of game descriptions to generate project files and instructions. This creates a surface for indirect prompt injection, which is handled through a structured decomposition process. 1. Ingestion points: User-provided game ideas in the initial prompt and clarifying questions. 2. Boundary markers: The process includes mandatory confirmation steps and genre classification before any file generation. 3. Capability inventory: The skill can write project files (TS, JSON, MD) and generate common development commands (npm). 4. Sanitization: Input is constrained by rigid templates and a fixed genre system, preventing open-ended instruction execution.
- [COMMAND_EXECUTION]: The skill generated instructions for standard development commands like 'npm install', 'npm run dev', and 'npm test'. These operations are performed locally on standard development tools and do not involve suspicious parameters or remote script piping.
Audit Metadata