The Agent Skills Directory

[PROMPT_INJECTION] (MEDIUM): The skill is susceptible to Indirect Prompt Injection because it fetches component implementations and demo examples from external registries.
Ingestion points: Untrusted data enters the agent context through shadcn___view_items_in_registries and shadcn___get_item_examples_from_registries.
Boundary markers: Absent. There are no instructions to delimit or ignore instructions embedded within the registry data.
Capability inventory: The skill is designed to generate shell commands (npx shadcn@latest add) and functional React code based on the ingested data.
Sanitization: Absent. The skill does not perform validation or escaping of the fetched registry content before using it to generate outputs.
[COMMAND_EXECUTION] (LOW): The skill generates shell commands for the user to execute.
Evidence: The tool shadcn___get_add_command_for_items is used specifically to construct npx installation strings.
[EXTERNAL_DOWNLOADS] (LOW): The skill relies on and executes the shadcn CLI tool from npm.
Evidence: Recommends npx shadcn@latest init and npx shadcn@latest add for project management.

shadcn-management