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
- Providers
- 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
Solid Chat UI
createChatHook from @tanstack/ai-solid/ui binds your chat options and components once. Tool calls and approvals stay typed.
Each screen then renders the chat with one hook call. You supply every visible component. There is no default markup or style.
Note
Upgrading from @tanstack/ai-solid-ui? Read Upgrade to 1.0.
Quick setup#
- Install the packages:
npm i @tanstack/ai @tanstack/ai-solid @tanstack/ai-openaipnpm add @tanstack/ai @tanstack/ai-solid @tanstack/ai-openaiyarn add @tanstack/ai @tanstack/ai-solid @tanstack/ai-openaibun add @tanstack/ai @tanstack/ai-solid @tanstack/ai-openai- 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 src/chat.tsx:
import { fetchServerSentEvents } from '@tanstack/ai-solid'
import { createChatHook } from '@tanstack/ai-solid/ui'
const chatOptions = {
connection: fetchServerSentEvents('/api/chat'),
}
const { useAppChat } = createChatHook({
options: chatOptions,
components: {
layout: (props) => (
<section>
<props.Messages />
</section>
),
message: (props) => (
<article data-role={props.message.role}>
<props.Parts />
</article>
),
},
partsComponents: {
text: (props) => <p>{props.part.content}</p>,
fallback: () => null,
},
})
export function ChatScreen() {
const chat = useAppChat({ threadId: 'support-chat' })
return (
<main>
<chat.AppChat />
<form
onSubmit={(event) => {
event.preventDefault()
const field = event.currentTarget.elements.namedItem('message')
if (!(field instanceof HTMLInputElement)) return
const text = field.value.trim()
if (!text) return
field.value = ''
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.
- useAppChat({ threadId }) creates one chat instance. chat.AppChat 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.
Read props as props.part and props.Messages. Do not destructure them, or Solid stops tracking the values.
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 Solid.
Render a tool#
Add the tool to tools in chatOptions. Then register a component under the tool name in toolsComponents. part.input and part.output are typed from the tool schemas.
toolsComponents: {
getWeather: (props) =>
props.part.state === 'complete' ? (
<p>
{props.part.input?.city}: {props.part.output?.temperature}
</p>
) : (
<p>Looking up weather</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: (props) =>
props.interrupt?.status === 'pending' ? (
<button onClick={() => props.interrupt?.resolveInterrupt(true)}>
Delete {props.part.input?.path}
</button>
) : (
<p>{props.part.input?.path}</p>
),
},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 props.Interrupts in the layout:
const choosePlan = defineInterrupt({
id: 'choosePlan',
payloadSchema: z.object({ title: z.string() }),
responseSchema: z.string(),
})
// In createChatHook:
interruptsComponents: {
generic: {
choosePlan: (props) => (
<button onClick={() => props.interrupt.resolveInterrupt('pro')}>
{props.interrupt.payload?.title}
</button>
),
fallback: (props) => <p>{props.interrupt.reason}</p>,
},
},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. Call useChatContext() inside the widget. This also moves the composer into the layout: register it as input, then render props.Input.
import { createSignal } from 'solid-js'
import { fetchServerSentEvents } from '@tanstack/ai-solid'
import { createChatHook } from '@tanstack/ai-solid/ui'
const chatOptions = {
connection: fetchServerSentEvents('/api/chat'),
}
const { useAppChat, useChatContext } = createChatHook({
options: chatOptions,
components: {
input: () => {
const chat = useChatContext()
const [draft, setDraft] = createSignal('')
return (
<form
onSubmit={(event) => {
event.preventDefault()
void chat.sendMessage(draft())
setDraft('')
}}
>
<input
value={draft()}
disabled={chat.isLoading()}
onInput={(event) => setDraft(event.currentTarget.value)}
/>
<button type="submit">Send</button>
</form>
)
},
layout: (props) => (
<main>
<props.Messages />
<props.Input />
</main>
),
message: (props) => (
<article>
<props.Parts />
</article>
),
},
partsComponents: {
text: (props) => <p>{props.part.content}</p>,
fallback: () => null,
},
})
export function ChatScreen() {
const chat = useAppChat({ threadId: 'support-chat' })
return <chat.AppChat />
}In Solid, chat.messages, chat.isLoading, and chat.error are accessors. Call them, such as chat.messages().length. Call useChatContext() only inside chat.AppChat. A call outside that tree throws.
Render markdown#
Model replies arrive as markdown. Register TextPart from @tanstack/ai-solid/ui as the text part:
import { TextPart } from '@tanstack/ai-solid/ui'
partsComponents: {
text: (props) => <TextPart content={props.part.content} />,
fallback: () => null,
},TextPart uses TanStack Markdown. It escapes raw HTML and removes executable URLs. To color fenced code, read Highlight markdown code.
Type a widget in its own file#
Move a widget to its own file and type its props with ToolProps, PartProps, or InterruptProps. Pass typeof chatOptions and a name. Keep the props object so Solid can track it:
import { fetchServerSentEvents } from '@tanstack/ai-solid'
import { createChatHook, type ToolProps } from '@tanstack/ai-solid/ui'
import { toolDefinition } from '@tanstack/ai'
import { z } from 'zod'
const getWeather = toolDefinition({
name: 'getWeather',
description: 'Look up weather',
inputSchema: z.object({ city: z.string() }),
outputSchema: z.object({ temperature: z.number() }),
}).client()
const chatOptions = {
connection: fetchServerSentEvents('/api/chat'),
tools: [getWeather],
}
export function WeatherTool(
props: ToolProps<typeof chatOptions, 'getWeather'>,
) {
return <strong>{props.part.input?.city}</strong>
}
export const { useAppChat } = createChatHook({
options: chatOptions,
components: {
layout: (props) => <props.Messages />,
message: (props) => (
<article>
<props.Parts />
</article>
),
},
partsComponents: { fallback: () => null },
toolsComponents: { getWeather: WeatherTool },
})Put chatOptions in a shared module so every widget file can import its type.
You now have a typed chat UI in Solid. 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 Solid, layout gets Messages, Interrupts, and Queue, plus Input when you register input.
useAppChat() takes per-instance options: threadId (required), initialMessages, live, and forwardedProps.
Approvals and generic interrupts#
- A tool approval renders in the tool when you read props.interrupt.
- To show it in the list instead, register a component under interruptsComponents.tools and render props.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'.
List input before layout#
If input is an inline function expression written after layout, TypeScript cannot see that an input is registered. Then Input is missing from the layout props.
To prevent this, put input first, or use an arrow function or a named component. If it still slips through, <props.Input /> without a registered input warns once in development.
Widgets in other files#
Call createChatHookContexts() in its own module. Pass chatContext, partContext, and interruptContext to createChatHook under context. Widget files then import useChatContext from the contexts module, with no circular import.
Walk messages yourself#
createChatUI is the lower-level kit. The UI.Messages render callback gets an accessor:
<UI.Messages>{(messages) => <span>{messages().length}</span>}</UI.Messages>TextPart options#
- highlighter: a synchronous CodeHighlighter. Its output is trusted HTML, so use createTanStackMarkdownHighlighter or another highlighter that escapes source text.
- extensions: more TanStack Markdown extensions. The streaming extension is always on.
- class, userClass, assistantClass: classes for the wrapper. Pass role to apply the role class.
If you still pass remarkPlugins, rehypePlugins, disableDefaultPlugins, or components, read TextPart markdown.