artifacts-builder

Pass

Audited by Gen Agent Trust Hub on Mar 6, 2026

Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSPROMPT_INJECTION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The init-artifact.sh and bundle-artifact.sh scripts download numerous packages from the public NPM registry, including pnpm, vite@5.4.11, parcel, and tailwindcss@3.4.1. These are well-known development tools but constitute external code dependencies.
  • [EXTERNAL_DOWNLOADS]: The skill references and extracts a local archive shadcn-components.tar.gz during the project initialization phase to populate the source directory.
  • [COMMAND_EXECUTION]: The skill automates project setup by executing shell commands for global package management (npm install -g pnpm), file system modification (sed), and archive extraction (tar).
  • [COMMAND_EXECUTION]: Runtime execution of JavaScript via node -e is used to programmatically parse and update configuration files such as tsconfig.json and tsconfig.app.json.
  • [PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection (Category 8) as it is designed to process untrusted data into a final executable output.
  • Ingestion points: The project name argument passed to init-artifact.sh and user-modified source code within the project directory.
  • Boundary markers: No specific boundary markers or 'ignore' instructions are implemented in the resulting bundle.html to separate user-contributed content from system behavior.
  • Capability inventory: Full access to shell command execution, network access via package managers, and the ability to write to the file system.
  • Sanitization: No robust validation or sanitization is performed on user inputs (e.g., project names) before they are interpolated into configuration files or the HTML entry point.
Audit Metadata
Risk Level
SAFE
Analyzed
Mar 6, 2026, 06:38 AM