web-artifacts-builder

Pass

Audited by Gen Agent Trust Hub on Feb 17, 2026

Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [EXTERNAL_DOWNLOADS] (LOW): The skill installs over 50 Node.js packages from the npm registry across both init-artifact.sh and bundle-artifact.sh. While these are established libraries (Vite, Radix UI, React), the large dependency tree represents a standard supply chain risk.
  • [COMMAND_EXECUTION] (LOW): The script scripts/init-artifact.sh performs a global package installation (npm install -g pnpm). While common in development environments, global installations represent a form of environment persistence/modification. The severity is lowered as this is typical for the skill's primary build-tooling purpose.
  • [PROMPT_INJECTION] (LOW): Indirect Prompt Injection Surface (Category 8). The skill generates executable HTML/JS artifacts based on user-influenced content and suggests testing them via browser automation (Playwright/Puppeteer).
  • Ingestion points: scripts/init-artifact.sh takes a project name as input; the development phase involves generating code based on user prompts.
  • Boundary markers: None. The scripts do not use delimiters or sanitization when assembling the final bundle.html.
  • Capability inventory: File system modification (sed, cat, tar), dependency installation (pnpm), and code execution (node -e, parcel).
  • Sanitization: None observed. The tool relies on the integrity of the LLM's code generation.
Audit Metadata
Risk Level
SAFE
Analyzed
Feb 17, 2026, 06:23 PM