The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill retrieves component definitions from external shadcn registries via MCP tools, creating a surface where malicious registry data could influence agent actions.
Ingestion points: Data enters the context through mcp__shadcn__search_items_in_registries and mcp__shadcn__view_items_in_registries.
Boundary markers: None; there are no instructions for the agent to treat registry output as untrusted or to use delimiters for that content.
Capability inventory: The skill employs Bash for executing commands and Write for modifying the project's source code.
Sanitization: No evidence of automated validation or sanitization of data retrieved from the registry before it is used to generate commands or files.
[EXTERNAL_DOWNLOADS]: The skill uses npx shadcn@latest to download and execute code from the official shadcn registry, a well-known service in the frontend ecosystem.
[COMMAND_EXECUTION]: The skill instructs the agent to run shell commands using Bash for tasks such as component installation and running project tests via pnpm.

shadcn-components