Consolidate host styling patterns into unified documentation

Merge the separate "Support native host styling using CSS variables" and
"Reacting to light/dark theme changes" sections into a single "Matching
host styling (CSS variables, theme, and fonts)" pattern. This provides
clearer guidance by covering all three concerns together.

- Rename `docs/patterns.ts` to `docs/patterns.tsx` for JSX support
- Add `hostStylingVanillaJs` and `hostStylingReact` consolidated examples
- Enhance `applyHostFonts_fromHostContext` to apply initial fonts after
  `connect()`, not just on context change
- Add CSS usage examples to `applyHostStyleVariables()` JSDoc
- Update font example to use `var(--font-sans, system-ui, sans-serif)`

Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
This commit is contained in:
Jonathan Hefner 2026-01-19 13:19:36 -06:00
commit 566ac58ab7
2 changed files with 108 additions and 26 deletions

View file

@ -18,49 +18,41 @@ Set {@link types!McpUiToolMeta.visibility Tool.\_meta.ui.visibility} to `["app"]
- See [PDF example](https://github.com/modelcontextprotocol/ext-apps/blob/main/examples/pdf-viewer) to read binaries by chunks to avoid call tool size limitations on platforms like claude.ai
- Pass auth token in `_meta` (will be loaded again in the future) + refresh token + store in local storage (see Persist data section below)
{@includeCode ./patterns.ts#authenticatedCalls}
{@includeCode ./patterns.tsx#authenticatedCalls}
## [TODO] Giving errors back to model
- Before app runs: validate inputs in tool call
- After it runs: use `updateModelContext`
{@includeCode ./patterns.ts#errorsToModel}
{@includeCode ./patterns.tsx#errorsToModel}
## Support native host styling using CSS variables
## Matching host styling (CSS variables, theme, and fonts)
Hosts provide CSS variables (e.g., `--color-background-primary`) that match their theme. Apply these to the document, then use them in your styles via `var()`.
Use the SDK's style helpers to apply host styling, then reference them in your CSS:
- **CSS variables** — Use `var(--color-background-primary)`, etc. in your CSS
- **Theme** — Use `[data-theme="dark"]` selectors or `light-dark()` function for theme-aware styles
- **Fonts** — Use `var(--font-sans)` or `var(--font-mono)` with fallbacks (e.g., `font-family: var(--font-sans, system-ui, sans-serif)`)
**Vanilla JS:**
{@includeCode ../src/styles.examples.ts#applyHostStyleVariables_fromHostContext}
{@includeCode ./patterns.tsx#hostStylingVanillaJs}
**React:**
{@includeCode ../src/react/useHostStyles.examples.tsx#useHostStyles_basicUsage}
## Reacting to light/dark theme changes
The host provides a `theme` value (`"light"` or `"dark"`) in the context. Apply it to `<html data-theme="...">` so you can use CSS selectors like `[data-theme="dark"]` for styling.
**Vanilla JS:**
{@includeCode ../src/styles.examples.ts#applyDocumentTheme_fromHostContext}
**React:** `useHostStyles()` (shown above) applies theme automatically. To read the theme reactively for conditional rendering:
{@includeCode ../src/react/useDocumentTheme.examples.tsx#useDocumentTheme_conditionalRender}
{@includeCode ./patterns.tsx#hostStylingReact}
## [TODO] Support fullscreen / exit fullscreen
{@includeCode ./patterns.ts#fullscreen}
{@includeCode ./patterns.tsx#fullscreen}
## [TODO] Persist data (incl. widget state)
- Note: OAI's `window.openai.setWidgetState({modelContent, privateContent, imageIds})` has only a partial equivalent in the MCP Apps spec (for now!): `App.updateModelContext({content, structuredContent})`
- For data persistence / to reload when conversation is reloaded, you must use localStorage / IndexedDb with `hostInfo.toolInfo.id` as key returned `CallToolResult._meta.widgetUUID = randomUUID()`
{@includeCode ./patterns.ts#persistData}
{@includeCode ./patterns.tsx#persistData}
## [TODO] Lower perceived latency / manage loading time
@ -68,13 +60,13 @@ Leverage partial inputs to show widgets as possible.
Beware of partial JSON being partial (but healed), so some of your objects may not be complete (e.g. in a list you may need to skip the last item if your code validates input schemas strictly).
{@includeCode ./patterns.ts#lowerPerceivedLatency}
{@includeCode ./patterns.tsx#lowerPerceivedLatency}
## [TODO] Supporting both iframe & MCP Apps in same binary
See recipe: https://github.com/modelcontextprotocol/ext-apps/issues/34
{@includeCode ./patterns.ts#iframeAndMcpApps}
{@includeCode ./patterns.tsx#iframeAndMcpApps}
## [TODO] Migrating from OpenAI to MCP Apps
@ -82,4 +74,4 @@ See [OpenAI -> MCP Apps](https://docs.google.com/document/d/13ROImOR9B8xc32yhqsF
Also: [Managing State](https://platform.openai.com/docs/actions/managing-state) (OpenAI)
{@includeCode ./patterns.ts#migrateFromOpenai}
{@includeCode ./patterns.tsx#migrateFromOpenai}

162
docs/patterns.tsx Normal file
View file

@ -0,0 +1,162 @@
/**
* Type-checked code examples for the patterns documentation.
*
* These examples are included in {@link ./patterns.md} via `@includeCode` tags.
* Each function's region markers define the code snippet that appears in the docs.
*
* @module
*/
import { App } from "../src/app.js";
import {
applyDocumentTheme,
applyHostFonts,
applyHostStyleVariables,
} from "../src/styles.js";
import type { McpUiHostContext } from "../src/types.js";
import { useApp, useHostStyles } from "../src/react/index.js";
/**
* Example: Authenticated calls from App
*/
function authenticatedCalls(app: App) {
//#region authenticatedCalls
// TODO: Use tool calls / read resources
// See PDF example to read binaries by chunks
// Pass auth token in _meta + refresh token + store in local storage
//#endregion authenticatedCalls
}
/**
* Example: Giving errors back to model
*/
function errorsToModel(app: App) {
//#region errorsToModel
// Before app runs: validate inputs in tool call
// After it runs: updateModelContext
// TODO: Complete implementation
//#endregion errorsToModel
}
/**
* Example: Unified host styling (theme, CSS variables, fonts)
*/
function hostStylingVanillaJs(app: App) {
//#region hostStylingVanillaJs
function applyHostContext(ctx: McpUiHostContext) {
if (ctx.theme) {
applyDocumentTheme(ctx.theme);
}
if (ctx.styles?.variables) {
applyHostStyleVariables(ctx.styles.variables);
}
if (ctx.styles?.css?.fonts) {
applyHostFonts(ctx.styles.css.fonts);
}
}
// Apply when host context changes
app.onhostcontextchanged = applyHostContext;
// Apply initial styles after connecting
app.connect().then(() => {
const ctx = app.getHostContext();
if (ctx) {
applyHostContext(ctx);
}
});
//#endregion hostStylingVanillaJs
}
/**
* Example: Host styling with React (CSS variables, theme, fonts)
*/
function hostStylingReact() {
//#region hostStylingReact
function MyApp() {
const { app } = useApp({
appInfo: { name: "MyApp", version: "1.0.0" },
capabilities: {},
});
// Apply all host styles (variables, theme, fonts)
useHostStyles(app, app?.getHostContext());
return (
<div
style={{
background: "var(--color-background-primary)",
fontFamily: "var(--font-sans)",
}}
>
<p>Styled with host CSS variables and fonts</p>
<p className="theme-aware">Uses [data-theme] selectors</p>
</div>
);
}
//#endregion hostStylingReact
}
/**
* Example: Support fullscreen / exit fullscreen
*/
function fullscreen() {
//#region fullscreen
// TODO: Implement fullscreen support
//#endregion fullscreen
}
/**
* Example: Persist data (incl. widget state)
*/
function persistData(app: App) {
//#region persistData
// Note: OAI's window.openai.setWidgetState({modelContent, privateContent, imageIds})
// has only a partial equivalent in MCP Apps: App.updateModelContext({content, structuredContent})
// For data persistence / to reload when conversation is reloaded,
// use localStorage / IndexedDb with hostInfo.toolInfo.id as key
// returned CallToolResult._meta.widgetUUID = randomUUID()
// TODO: Complete implementation
//#endregion persistData
}
/**
* Example: Lower perceived latency / manage loading time
*/
function lowerPerceivedLatency(app: App) {
//#region lowerPerceivedLatency
// TODO: Leverage partial inputs to show widgets as possible.
// Beware of partial JSON being partial (but healed),
// so some objects may not be complete.
//#endregion lowerPerceivedLatency
}
/**
* Example: Supporting both iframe & MCP Apps in same binary
*/
function iframeAndMcpApps() {
//#region iframeAndMcpApps
// TODO: See recipe: https://github.com/modelcontextprotocol/ext-apps/issues/34
//#endregion iframeAndMcpApps
}
/**
* Example: Migrating from OpenAI to MCP Apps
*/
function migrateFromOpenai() {
//#region migrateFromOpenai
// TODO: See OpenAI -> MCP Apps migration guide
// https://docs.google.com/document/d/13ROImOR9B8xc32yhqsFyC9Hh3_H63JFORDIyjyIPcU4/edit
//#endregion migrateFromOpenai
}
// Suppress unused variable warnings
void authenticatedCalls;
void errorsToModel;
void hostStylingVanillaJs;
void hostStylingReact;
void fullscreen;
void persistData;
void lowerPerceivedLatency;
void iframeAndMcpApps;
void migrateFromOpenai;