The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill relies on the @21st-dev/cli package, which is the official command-line interface provided by the vendor (21st-dev). All external references point to the legitimate vendor domain at 21st.dev.
[COMMAND_EXECUTION]: The instructions direct the agent to interact with the system using the 21st CLI binary. These operations (publish, edit, delete, profile management) are consistent with the skill's stated purpose of registry management.
[DATA_EXFILTRATION]: The skill facilitates the uploading of local code and configuration files to the 21st.dev registry. This is a core feature of the skill. To mitigate risk, the author includes a strict rule for agents: 'Never publish a file with API keys, secrets, or unsaved edits'.
[CREDENTIALS_UNSAFE]: The skill provides instructions for authenticating with the registry using API keys (e.g., 21st_sk_...) passed via command flags or environment variables (TWENTYFIRST_TOKEN, API_KEY_21ST). No credentials are hardcoded; the skill directs users to their official dashboard to generate their own keys.
[PROMPT_INJECTION]: The skill has a surface for indirect prompt injection as it processes local files and profile data which could contain malicious instructions.
Ingestion points: Reads local source files (Component.tsx, CSS themes) and JSON profile data via 21st profile get.
Boundary markers: The skill does not define specific delimiters for processed file content in its instructions.
Capability inventory: Perform network uploads and execute CLI commands.
Sanitization: There are no specific instructions for sanitizing or validating the contents of the files before they are processed by the CLI tool.

21st-registry