---
name: "@inferencesh"
slug: "inferencesh"
description: "A batteries-included shadcn registry from inference.sh providing a complete AI agent UI runtime. Key blocks: Agent (full wired agent with streaming and tool lifecycle), Chat (container, messages, composer, status indicators), Tools (tool invocation UI: pending, in-progress, human-in-the-loop approval, results), Widgets (renders declarative JSON UI nodes as interactive components), Markdown (high-fidelity with code blocks and zoomable images), Code Block, Steps, Sidebar Light, Table of Contents, YouTube Embed, and Zoomable Image. Open-source, MIT. Use for: building production AI agent interfaces in Next.js that need streaming, tool rendering, and human-in-the-loop approvals without wiring each piece from scratch."
category: "AI/chat UI"
tags: ["ai", "blocks", "dashboard", "developer-tools", "loaders", "media", "tables"]
framework: "react"
license: null
author: "inference-sh"
website: "https://ui.inference.sh"
github: "https://github.com/inference-sh/shadcn-registry"
registry: "https://ui.inference.sh/r/registry.json"
pro: false
pro_url: null
github_stars: 21
component_count: 14
generated_at: "2026-07-28T08:59:05.895Z"
---

# @inferencesh

A batteries-included shadcn registry from inference.sh providing a complete AI agent UI runtime. Key blocks: Agent (full wired agent with streaming and tool lifecycle), Chat (container, messages, composer, status indicators), Tools (tool invocation UI: pending, in-progress, human-in-the-loop approval, results), Widgets (renders declarative JSON UI nodes as interactive components), Markdown (high-fidelity with code blocks and zoomable images), Code Block, Steps, Sidebar Light, Table of Contents, YouTube Embed, and Zoomable Image. Open-source, MIT. Use for: building production AI agent interfaces in Next.js that need streaming, tool rendering, and human-in-the-loop approvals without wiring each piece from scratch.

## Links

- Website: https://ui.inference.sh
- GitHub: https://github.com/inference-sh/shadcn-registry
- Registry JSON: https://ui.inference.sh/r/registry.json
- Directory page: https://sh4dcn.vercel.app/registry/inferencesh
- This file: https://sh4dcn.vercel.app/md/inferencesh.md

## Install

```sh
# add the whole registry (shadcn CLI)
npx shadcn@latest add https://ui.inference.sh/r/registry.json
pnpm dlx shadcn@latest add https://ui.inference.sh/r/registry.json
bunx shadcn@latest add https://ui.inference.sh/r/registry.json
# clone the source repo
git clone https://github.com/inference-sh/shadcn-registry.git
# fetch the raw registry index
curl -s https://ui.inference.sh/r/registry.json
```

## Components (14)

| name | type | description | install |
| --- | --- | --- | --- |
| agent | registry:block | Full-featured AI agent chat with state management, tools, and widgets. | `npx shadcn@latest add https://ui.inference.sh/r/agent.json` |
| chat | registry:block | Virtualized chat with measured markdown. Auto-scroll, message strategies, typing indicators. | `npx shadcn@latest add https://ui.inference.sh/r/chat.json` |
| tools | registry:block | Displays tool lifecycle: pending, in-progress, approval, and results. | `npx shadcn@latest add https://ui.inference.sh/r/tools.json` |
| task | registry:block | Real-time task display with streaming support, status indicators, and logs viewer. | `npx shadcn@latest add https://ui.inference.sh/r/task.json` |
| widgets | registry:block | Renders declarative UI nodes (forms, buttons, cards) from agent tool outputs. | `npx shadcn@latest add https://ui.inference.sh/r/widgets.json` |
| pretext-md | registry:block | Measured markdown renderer powered by pretext. DOM-free text measurement, plugin architecture, virtualizable. | `npx shadcn@latest add https://ui.inference.sh/r/pretext-md.json` |
| virtualize | registry:block | Typed virtualization with measurement strategies, incremental caching, and scroll correction. | `npx shadcn@latest add https://ui.inference.sh/r/virtualize.json` |
| markdown | registry:block | react-markdown based renderer. Consider using pretext-md instead. | `npx shadcn@latest add https://ui.inference.sh/r/markdown.json` |
| code-block | registry:block | Syntax-highlighted code block with line numbers and copy button. | `npx shadcn@latest add https://ui.inference.sh/r/code-block.json` |
| youtube-embed | registry:block | Responsive YouTube video embed with loading state. | `npx shadcn@latest add https://ui.inference.sh/r/youtube-embed.json` |
| zoomable-image | registry:block | Click-to-zoom image component with medium.com-style lightbox. | `npx shadcn@latest add https://ui.inference.sh/r/zoomable-image.json` |
| steps | registry:block | Numbered step component with connecting line for sequential instructions. | `npx shadcn@latest add https://ui.inference.sh/r/steps.json` |
| sidebar-light | registry:block | Lightweight sidebar navigation with nested items and active state. | `npx shadcn@latest add https://ui.inference.sh/r/sidebar-light.json` |
| table-of-contents | registry:block | Auto-scrolling table of contents with intersection observer for active section. | `npx shadcn@latest add https://ui.inference.sh/r/table-of-contents.json` |

## For AI agents

```text
You are adding components from @inferencesh to a shadcn-compatible project.
1. Read this file (https://sh4dcn.vercel.app/md/inferencesh.md) for the component list.
2. Fetch the machine-readable index: https://ui.inference.sh/r/registry.json
3. Install a single component with its install command from the table above.
4. Full structured record: https://sh4dcn.vercel.app/json (library slug: inferencesh).
```