web-artifacts-builder

Pass

Audited by Gen Agent Trust Hub on Sep 19, 2026

Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONPRIVILEGE_ESCALATIONINDIRECT_PROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill utilizes shell scripts (init-artifact.sh and bundle-artifact.sh) to automate the React development workflow, including project creation via Vite, source code modification, and production bundling using Parcel.
  • [EXTERNAL_DOWNLOADS]: During initialization and bundling, the scripts download standard and well-known frontend packages from the NPM registry, such as vite, tailwindcss, radix-ui, and parcel. These are standard tools for modern web development.
  • [PRIVILEGE_ESCALATION]: The initialization script (init-artifact.sh) includes a check for the pnpm package manager and attempts to install it globally (npm install -g pnpm) if it is not found in the environment.
  • [INDIRECT_PROMPT_INJECTION]: As a tool for building artifacts from user instructions, this skill possesses an inherent vulnerability surface. User-provided content used to generate the artifact's code could contain malicious instructions or scripts (XSS).
  • Ingestion points: User-provided requirements and content for the artifact (SKILL.md).
  • Boundary markers: None explicitly defined for user input within the scripts.
  • Capability inventory: The skill has the capability to write files and execute build commands to generate a functional HTML/JS artifact.
  • Sanitization: None present in the infrastructure; safety relies on the agent's internal content filtering during the code generation phase.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 19, 2026, 12:32 PM