The Agent Skills Directory

[DYNAMIC_CONTEXT_INJECTION]: The skill uses the ! syntax in SKILL.md to execute npx fabricator-ui@latest info --json at load time. This command gathers project information such as the framework, installed components, and Tailwind configuration, allowing the agent to provide context-aware assistance.
[COMMAND_EXECUTION]: The skill provides instructions for the agent to use the fabricator-ui CLI via package runners like npx, pnpm dlx, yarn dlx, and bunx. These commands are used for searching, adding, and auditing UI components within the user's project.
[EXTERNAL_DOWNLOADS]: The skill fetches component source code, metadata, and documentation from https://fabricator-ui.vercel.app. These downloads are directed to the official domain of the Fabricator UI project and are necessary for the skill's functionality.
[INDIRECT_PROMPT_INJECTION]: The skill possesses an attack surface for indirect prompt injection through data ingestion.
Ingestion points: Project metadata from npx info (SKILL.md), component documentation from the Fabricator website (SKILL.md), and item JSON from the registry (registry.md).
Boundary markers: The skill does not currently use specific delimiters or instructions to ignore potential commands embedded in registry data.
Capability inventory: The skill allows for file writing through the add command, shell command execution via the Bash tool, and network operations for fetching documentation.
Sanitization: The skill relies on standard LLM safety guardrails as no explicit sanitization of registry content is documented.

fabricator