The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill facilitates the download of UI component configurations and registry files from the vendor's official domain at ui.inference.sh. These resources are used to scaffold the agent interface.
[COMMAND_EXECUTION]: Documentation includes standard commands for installing the @inferencesh/sdk via npm and adding UI building blocks (chat-ui, widgets-ui, tools-ui) using the npx skills and shadcn CLIs.
[PROMPT_INJECTION]: The skill documents an agent component that processes instructions from an LLM to trigger client-side tools (e.g., scan_ui, fill_field) and render generative UI widgets. This architecture creates an indirect prompt injection surface where untrusted data processed by the LLM could attempt to influence the agent's UI actions.
Ingestion points: The Agent component consumes dynamic responses from an LLM via a defined proxyUrl (SKILL.md).
Boundary markers: No explicit delimiter or sanitization logic is shown in the provided usage examples to differentiate between LLM instructions and data.
Capability inventory: Supports client-side tool execution, form filling, and human-in-the-loop approval workflows.
Sanitization: Not visible in the basic implementation snippets provided in the documentation.

agent-ui