atai-design-system

Pass

Audited by Gen Agent Trust Hub on Jul 22, 2026

Risk Level: SAFEREMOTE_CODE_EXECUTIONCOMMAND_EXECUTIONEXTERNAL_DOWNLOADS
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The skill instructs the agent to install and update various npm packages under the @archetypeai/ scope and the layerchart library from the npm registry.
  • [REMOTE_CODE_EXECUTION]: The skill uses npx to execute the vendor's command-line interface tool (@archetypeai/ds-cli) and the well-known shadcn-svelte utility for project scaffolding and component management.
  • [COMMAND_EXECUTION]: Shell commands are used to create project directories, initialize configurations, and modify local files such as svelte.config.js and .gitignore.
  • [EXTERNAL_DOWNLOADS]: The ds add workflow involves fetching component source files from vendor-hosted registries located at design-system-labs.archetypeai.workers.dev and design-system-console.archetypeai.workers.dev.
  • [PROMPT_INJECTION]: The skill designates ds-manifest.json as the machine-readable source of truth for component logic, which represents an indirect prompt injection surface where external data influences agent instructions. This is mitigated by the data originating from the vendor's own scaffolding process.
Audit Metadata
Risk Level
SAFE
Analyzed
Jul 22, 2026, 04:58 PM
Security Audit — agent-trust-hub — atai-design-system