The Agent Skills Directory

[PROMPT_INJECTION]: The skill introduces a surface for indirect prompt injection by fetching instructions from an external API.
Ingestion points: The skill fetches aiRules and other metadata from the StyleKit API (https://www.stylekit.top/api/styles/{slug}) at runtime.
Boundary markers: None identified; instructions explicitly tell the agent to treat fetched aiRules with "Highest priority," which could override other agent constraints.
Capability inventory: The agent is directed to execute shell commands (npx shadcn add) and modify project files based on the fetched data.
Sanitization: There is no mention of sanitization or validation of the fetched remote instructions before they are processed by the agent.
[REMOTE_CODE_EXECUTION]: The skill facilitates the installation of remote components via npx shadcn add https://www.stylekit.top/r/<slug>.json. This command uses the shadcn tool to download and integrate code from the vendor's domain into the user's project environment.
[COMMAND_EXECUTION]: The instructions guide the agent to perform setup and styling tasks using CLI tools such as npx and shadcn.
[EXTERNAL_DOWNLOADS]: The skill fetches design specifications, tokens, and component recipes from the stylekit.top API to guide the UI generation process.

stylekit