atai-design-system

Pass

Audited by Gen Agent Trust Hub on Aug 18, 2026

Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONREMOTE_CODE_EXECUTION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The skill fetches design system tokens and primitives from official Archetype AI npm packages and vendor-hosted component registries on Cloudflare Workers. All external sources are verified as belonging to the skill author's infrastructure.
  • [COMMAND_EXECUTION]: The agent is instructed to use npx to execute the @archetypeai/ds-cli and shadcn-svelte tools for project initialization and component management. These commands are standard for modern web development workflows.
  • [REMOTE_CODE_EXECUTION]: Executing the @latest version of the design system CLI via npx involves running remote code. This is a trusted operation as the tool is part of the vendor's official design system distribution.
  • [SAFE]: The skill provides explicit guidance on project security, recommending the use of .gitignore for commercial font files to prevent their inclusion in public repositories.
  • [SAFE]: The skill implements a multi-step configuration chain where the CLI installs local CLAUDE.md and ds-manifest.json files as the project's source of truth. Ingestion points: The agent reads these files from the project root after scaffolding. Boundary markers: The skill establishes these files as the authoritative source for future component usage. Capability inventory: The agent uses manifest recipes to generate UI markup. Sanitization: None; the process relies on the established trust of the vendor-provided scaffolding tools.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 18, 2026, 09:54 PM
Security Audit — agent-trust-hub — atai-design-system