The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: Fetches and installs UI components, themes, and templates from the 21st.dev registry. It also utilizes the @21st-dev/cli package and shadcn for installation processes.
[COMMAND_EXECUTION]: Executes shell commands via the 21st CLI and npx for searching, pulling code, and installing components. Commands like 21st init --write and 21st install-skill perform modifications to the local environment and configuration files.
[DATA_EXFILTRATION]: Accesses sensitive local configuration paths (~/.config/21st/auth.json) to store and retrieve authentication tokens. It also prompts the use of sensitive environment variables like TWENTYFIRST_TOKEN and API_KEY_21ST.
[INDIRECT_PROMPT_INJECTION]:
Ingestion points: Reads search results, component code, and theme CSS from the external 21st.dev catalog (SKILL.md).
Capability inventory: Can write files to the project directory (components/ui/), install NPM dependencies, and execute shell commands through the CLI (SKILL.md).
Boundary markers: The skill does not define specific delimiters or instructions for the agent to ignore potentially malicious instructions embedded in the downloaded component code or metadata.
Sanitization: Relies on the integrity of the vendor's registry and the standard shadcn component resolution logic.

21st-cli-use