skills/21st-dev/skill/21st-cli/Gen Agent Trust Hub

21st-cli

Pass

Audited by Gen Agent Trust Hub on Jul 3, 2026

Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADS
Full Analysis
  • [COMMAND_EXECUTION]: The skill uses shell commands via 21st and npx to manage UI components, such as 21st add for installation and 21st publish for uploading local code.
  • [EXTERNAL_DOWNLOADS]: The skill instructs the agent to install the @21st-dev/cli package and associated dependencies from the npm registry and fetches component code from the official 21st.dev registry.
  • [CREDENTIALS_UNSAFE]: The skill manages authentication by reading from ~/.config/21st/auth.json and supports passing API keys via standard environment variables like TWENTYFIRST_TOKEN or API_KEY_21ST.
  • [INDIRECT_PROMPT_INJECTION]: The skill possesses an indirect prompt injection surface as it downloads external React component code (21st add) and processes user-provided descriptions for AI generation (21st generate). While no explicit sanitization or boundary markers are documented, the capabilities are restricted to the local development environment and the 21st.dev platform.
Audit Metadata
Risk Level
SAFE
Analyzed
Jul 3, 2026, 10:11 AM
Security Audit — agent-trust-hub — 21st-cli