The Agent Skills Directory

[REMOTE_CODE_EXECUTION]: The skill instructs the agent to execute remote tools using npx, specifically npx shadcn@latest mcp and @jpisnice/shadcn-ui-mcp-server. While the official shadcn CLI is a well-known service, the community-contributed MCP server represents an external dependency executed at runtime.
[COMMAND_EXECUTION]: The skill performs several shell-level operations, including project initialization with npx shadcn@latest init, adding components with npx shadcn add, and auditing the codebase using grep -rc 'style={{' src/ to count inline style occurrences.
[INDIRECT_PROMPT_INJECTION]: The skill has an attack surface for indirect prompt injection as it ingests untrusted data from project files and external component registries.
Ingestion points: Reads the project's components.json file and component source code/metadata from ui.shadcn.com and prompt-kit.com via MCP tools.
Boundary markers: None explicitly defined in the instructions to prevent the agent from following instructions embedded in the external component documentation or source code.
Capability inventory: The skill has the capability to execute shell commands (npx, grep) and read/write project files through its configured MCP servers.
Sanitization: There is no evidence of sanitization or validation of the content retrieved from external registries before the agent processes it for UI generation.

shadcn-first