The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill facilitates fetching component source code and UI blocks from external registries using MCP tools and web_fetch. This creates a surface for indirect prompt injection if the external data contains malicious instructions meant to influence the agent.
Ingestion points: Component metadata, source code, and blocks retrieved from registries as specified in SKILL.md.
Boundary markers: Absent; the instructions do not implement delimiters or explicit 'ignore instructions' warnings for content fetched from registries.
Capability inventory: The skill possesses the capability to Write files locally and execute Bash commands for project setup.
Sanitization: Absent; the skill does not explicitly validate or sanitize the textual content of fetched components before processing.
[COMMAND_EXECUTION]: The skill utilizes Bash to execute standard project management commands such as npx shadcn for component installation and scripts/verify-setup.sh for configuration validation. These commands are typical for the skill's intended developer use case.
[EXTERNAL_DOWNLOADS]: The skill references and executes tools from trusted organizations and well-known services. This includes fetching code from the shadcn/ui registry and installing global skills from repositories associated with trusted vendors.

shadcn-ui