What problem does it solve? Building interfaces that display AI agent tool calls requires handling multiple lifecycle states like pending, running, approval, success, and error, which is tedious to implement from scratch in React or Next.js applications. ## Core Features & Use Cases - Tool Call Display: Render tool invocations with name, arguments, and live status using the ToolCall component. - Human-in-the-Loop Approvals: Show approval gates with approve and deny callbacks via the ToolApproval component. - Result Rendering: Display completed tool outputs with the ToolResult component, including automatic icons based on tool name patterns. - Use Case: When building a chat interface for an AI agent that calls tools like search_web or send_email, use these components to show users what the agent is doing and let them approve sensitive actions before execution. ## Quick Start Install the tool UI components into my Next.js project and render a tool call with approval flow for a send_email function.