Persistence (Advanced)
- Build your first harness
- Host options
- Send and show media
- Connect clients
- Write a client in any language
- Access and tenants
- Run in the terminal
- Deploy a harness
- Send inputs safely
- Run a tool without a model
- Durable sessions
- Run background work on workers
- List, rename, and fork sessions
- Lifecycle and retention
- Run side effects once
- Wait for a person
- Build on the session log
- Control how a turn ends
- Store settings per thread
- Compact a long session
- Fork and reset a thread
- One log for a session tree
- Write a plugin
- Track usage and cost
- Ask before risky tool calls
- Give your agent coding tools
- Sandbox and extend the coding tools
- Run the coding tools for many threads
- Undo a turn
- Build a coding agent
- Delegate to coding agents
- Switch between agent profiles
- Auth and connectors
- Share a thread between people
- Connect model providers
- Build your own UI
- Code mode in a harness
- Work until a goal is met
- Run agents from a harness
- Hand off a conversation
- Add skills
- Use MCP servers
- Give your agent a computer
- Use from any MCP client
- Use a harness from an editor (ACP)
- Overview
- Quick Start
- ProvidersUpdated
- Harnesses
- Harness Auth
- Workspace
- Tools
- Policy
- Lifecycle & Snapshots
- Portable Snapshots
- Keep Files After Reload
- Save a Named Version
- Branch From a Version
- Send a Frozen File
- Let the Agent Save and Fork
- Pick Which Files to Keep
- What a Snapshot Stores
- Instance Durability
- Durable Runs
- Events
Sandboxes (Advanced)
- Runtime Context
- Bring Your Own Key
- Keep Keys on the Server
- Debug Logging
- Multimodal Content
- Files API
- Per-Model Type Safety
- Runtime Adapter Switching
- Tree-Shaking
- Extend Adapter
- Extend Adapter: Advanced
- Typed Pre-Configured Options
- Test with a Fake Model
- Prompt Caching
- Mid-Conversation Changes
- Approval Flow Processing
Remix Chat UI
createChatHook from @tanstack/ai-remix/ui binds your chat options and components once. Tool calls and approvals stay typed.
Each screen then creates a chat with createAppChat(handle, { threadId }) and renders it with ui.Chat. Each widget is a Remix setup function that reads handle.props. You supply every visible component. There is no default markup or style.
Quick setup#
- Install the packages:
npm i @tanstack/ai @tanstack/ai-remix @tanstack/ai-openai remixpnpm add @tanstack/ai @tanstack/ai-remix @tanstack/ai-openai remixyarn add @tanstack/ai @tanstack/ai-remix @tanstack/ai-openai remixbun add @tanstack/ai @tanstack/ai-remix @tanstack/ai-openai remix- Put OPENAI_API_KEY in your server environment. Then add a POST route at /api/chat:
import {
chat,
chatParamsFromRequest,
toServerSentEventsResponse,
} from '@tanstack/ai'
import { openaiText } from '@tanstack/ai-openai'
export async function POST(request: Request) {
const { messages, threadId, runId } = await chatParamsFromRequest(request)
const stream = chat({
adapter: openaiText('gpt-6.1-sol'),
messages,
threadId,
runId,
})
return toServerSentEventsResponse(stream)
}- Create app/chat.tsx:
import { fetchServerSentEvents } from '@tanstack/ai-remix'
import {
createChatHook,
type LayoutProps,
type MessageProps,
type PartProps,
} from '@tanstack/ai-remix/ui'
import { clientEntry, on, type Handle } from 'remix/ui'
const chatOptions = {
connection: fetchServerSentEvents('/api/chat'),
}
const { createAppChat, ui } = createChatHook({
options: chatOptions,
components: {
layout(handle: Handle<LayoutProps<typeof chatOptions>>) {
return () => {
const { Messages } = handle.props
return (
<section>
<Messages />
</section>
)
}
},
message(handle: Handle<MessageProps<typeof chatOptions>>) {
return () => {
const { message, Parts } = handle.props
return (
<article data-role={message.role}>
<Parts />
</article>
)
}
},
},
partsComponents: {
text(handle: Handle<PartProps<typeof chatOptions, 'text'>>) {
return () => <p>{handle.props.part.content}</p>
},
fallback() {
return () => null
},
},
})
export const ChatScreen = clientEntry(
import.meta.url,
function ChatScreen(handle: Handle) {
const chat = createAppChat(handle, { threadId: 'support-chat' })
return () => (
<main>
<ui.Chat chat={chat} />
<form
mix={on('submit', (event) => {
event.preventDefault()
const form = event.currentTarget
const text = String(new FormData(form).get('message') ?? '').trim()
if (text === '') {
return
}
form.reset()
void chat.sendMessage(text)
})}
>
<input name="message" />
<button type="submit">Send</button>
</form>
</main>
)
},
)- Render <ChatScreen /> and send a message.
The reply streams into the page as the model writes it.
How it works#
- createChatHook runs once, at module scope. It binds chatOptions to your components. It returns ui, the kit with ui.Chat.
- createAppChat(handle, { threadId }) creates one chat instance inside a setup function. ui.Chat renders that instance with your components.
- layout places the message list. message wraps one message. Parts picks a component for each part of that message.
- partsComponents.text renders text. fallback catches every part type that you did not register.
Render Messages and Parts as tags, such as <Messages />, not as function calls.
Next steps#
Each step adds one thing to the quick setup. The chat UI recipes have the full file for each one. The recipes use React, and the option names are the same in Remix.
Render a tool#
Add the tool to tools in chatOptions. Then register a component under the tool name in toolsComponents. Type its handle with ToolProps:
toolsComponents: {
getWeather(handle: Handle<ToolProps<typeof chatOptions, 'getWeather'>>) {
return () => {
const { part } = handle.props
if (part.state !== 'complete') return <p>Looking up weather</p>
return (
<p>
{part.input?.city}: {part.output?.temperature}
</p>
)
}
},
},The factory needs one entry for every tool in chatOptions. Full example: Make a tool look right.
Ask before a tool runs#
Set needsApproval: true on the tool. The tool component then gets an interrupt prop:
toolsComponents: {
deleteFile(handle: Handle<ToolProps<typeof chatOptions, 'deleteFile'>>) {
return () => {
const { part, interrupt } = handle.props
if (interrupt?.status !== 'pending') return <p>{part.input?.path}</p>
return (
<button
mix={on('click', () => interrupt.resolveInterrupt(true))}
>
Delete {part.input?.path}
</button>
)
}
},
},resolveInterrupt(true) approves. resolveInterrupt(false) denies. Full example: Ask before a tool runs.
Ask the user your own question#
Define the question with defineInterrupt. Add it to interrupts in chatOptions. Register a component under its id in interruptsComponents.generic, and render Interrupts in the layout:
const choosePlan = defineInterrupt({
id: 'choosePlan',
payloadSchema: z.object({ title: z.string() }),
responseSchema: z.string(),
})
// In createChatHook:
interruptsComponents: {
generic: {
choosePlan(
handle: Handle<InterruptProps<typeof chatOptions, 'choosePlan'>>,
) {
return () => {
const { interrupt } = handle.props
return (
<button mix={on('click', () => interrupt.resolveInterrupt('pro'))}>
{interrupt.payload?.title}
</button>
)
}
},
},
},The factory needs one entry for every interrupt id in chatOptions. Full example: Ask the user your own question.
Read chat state inside a widget#
Mapped components do not get chat as a prop. Import ui and call ui.useChatContext(handle) inside the widget. This also moves the composer into the layout. Create app/chat-input.tsx:
import { on, type Handle, type RemixNode } from 'remix/ui'
import { ui } from './chat'
export function ChatInput(handle: Handle): () => RemixNode {
const chat = ui.useChatContext(handle)
return () => (
<form
mix={on('submit', (event) => {
event.preventDefault()
const form = event.currentTarget
const text = String(new FormData(form).get('message') ?? '').trim()
if (text === '') {
return
}
form.reset()
void chat.sendMessage(text)
})}
>
<input name="message" disabled={chat.isLoading} />
<button type="submit">Send</button>
</form>
)
}Then:
- Export ui from app/chat.tsx.
- Register input: ChatInput on components.
- Render <Input /> in the layout, from handle.props.
- Remove the form from ChatScreen.
The return type on ChatInput stops a circular type between the widget and the ui that registers it. Call ui.useChatContext(handle) only under ui.Chat or ui.Provider. A call outside that tree throws.
Render markdown#
TextPart from @tanstack/ai-remix/ui renders the content as plain text, with no markdown. To render markdown, render part.content in your own text part with the markdown library of your choice.
Type a widget in its own file#
Type the handle.props of a widget with ToolProps, PartProps, InterruptProps, LayoutProps, or MessageProps. Pass typeof chatOptions and a name. Share chatOptions from its own module:
import type { ToolProps } from '@tanstack/ai-remix/ui'
import type { Handle } from 'remix/ui'
import { chatOptions } from './chat-options'
export function WeatherTool(
handle: Handle<ToolProps<typeof chatOptions, 'getWeather'>>,
) {
return () => <strong>{handle.props.part.input?.city}</strong>
}You now have a typed chat UI in Remix. Each widget is your own component.
Advanced#
Slots and props#
The config keys, part keys, and prop types are the same as in React. See Every widget slot. In Remix, layout gets Messages, Interrupts, Queue, and Input. If you render Input with no registered input, it renders nothing and warns once in development.
createAppChat(handle, overrides) takes per-instance options: threadId (required), initialMessages, live, and forwardedProps.
Approvals and generic interrupts#
- A tool approval renders in the tool when you read the interrupt prop.
- To show it in the list instead, register a component under interruptsComponents.tools and render Interrupts.
- Generic interrupts always render in the list. generic.fallback renders an unregistered id and an unbound interrupt. If the copy must be different, branch on interrupt.kind === 'unbound'.
Show queued messages#
Register queue on components, then render <Queue /> in the layout. A queue component gets one item. Call item.cancelQueued() to drop it.