21st-cli-use

HIGH W007: Insecure credential handling detected in skill instructions.

• Insecure credential handling detected (high risk: 0.80). The skill docs explicitly show an API key format ("21st_sk_…") and instruct passing it via a command-line flag (--api-key ), which encourages including secrets verbatim in generated commands/outputs (an exfiltration risk); environment-variable usage is mentioned as a safer alternative but the CLI-arg pattern is present.

MEDIUM W011: Third-party content exposure detected (indirect prompt injection risk).

• Third-party content exposure detected (high risk: 0.75). The required workflow is the 21st CLI’s “Find / Pull code / Install” flow, which fetches component/theme/template code from the 21st.dev catalog at runtime (public/outsider-authored registry content) and then injects that retrieved free-form code text into the agent/LLM context for generation/installation.

MEDIUM W012: Unverifiable external dependency detected (runtime URL that controls agent).

• Potentially malicious external URL detected (high risk: 0.90). The SKILL.md includes a runtime command that calls npx shadcn@latest add "https://21st.dev/r//" (line SKILL.md:77), which fetches and installs remote component code and therefore executes remote code at runtime (https://21st.dev/r//).