The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes a local shell script scripts/verify-setup.sh to validate the project's file structure, configuration files (like components.json and tailwind.config.js), and dependency status.
[EXTERNAL_DOWNLOADS]: The skill instructions and documentation guide the agent to use npx and web_fetch to download UI components, project boilerplates, and documentation from the shadcn registry and GitHub repositories.
[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it is designed to fetch and process component source code from external registries. If an attacker gains control over a registry, they could inject malicious instructions into the component code that the agent then processes while integrating the component into the user's project.
Ingestion points: Component source code retrieved via get_component and get_block MCP tools from registries specified in components.json.
Boundary markers: Absent; there are no specific instructions for the agent to ignore natural language commands embedded in the retrieved code.
Capability inventory: The agent has Bash, Write, and web_fetch capabilities, allowing it to modify files or execute commands based on injected instructions.
Sanitization: No explicit sanitization or filtering of the retrieved source code is performed before the agent processes it.

shadcn-ui