agui-author
Author live dashboard UI from an agent via the `emit_ui` MCP tool. Emit
This skill should be used when the user asks to "add MCP App support to my web app", "turn my web app into a hybrid MCP App", "make my web page work as an MCP App too", "wrap my existing UI as an MCP App", "convert iframe embed to MCP App", "turn my SPA into an MCP App", or
$ npx -y skills add awslabs/cli-agent-orchestrator --skill convert-web-app --agent claude-codeHow it fires
How this skill gets triggered: by you, by Claude, or both.
/convert-web-appContext preview
The summary Claude sees to decide when to auto-load this skill.
This skill should be used when the user asks to "add MCP App support to my web app", "turn my web app into a hybrid MCP App", "make my web page work as an MCP App too", "wrap my existing UI as an MCP App", "convert iframe embed to MCP App", "turn my SPA into an MCP App", or
name: convert-web-app description: This skill should be used when the user asks to "add MCP App support to my web app", "turn my web app into a hybrid MCP App", "make my web page work as an MCP App too", "wrap my existing UI as an MCP App", "convert iframe embed to MCP App", "turn my SPA into an MCP App", or needs to add MCP App support to an existing web application while keeping it working standalone. Provides guidance for analyzing existing web apps and creating a hybrid web + MCP App with server-side tool and resource registration.
Add MCP App support to an existing web application so it works both as a standalone web app **and** as an MCP App that renders inline in MCP-enabled hosts like Claude Desktop — from a single codebase.
The existing web app stays intact. A thin initialization layer detects whether the app is running inside an MCP host or as a regular web page, and fetches parameters from the appropriate source. A new MCP server wraps the app's bundled HTML as a resource and registers a tool to display it.
Standalone: Browser loads page → App reads URL params / APIs → renders MCP App: Host calls tool → Server returns result → Host renders app in iframe → App reads MCP lifecycle → renders
The app's rendering logic is shared — only the data source changes.
Clone the SDK repository for working examples and API documentation:
git clone --branch "v$(npm view @modelcontextprotocol/ext-apps version)" --depth 1 https://github.com/modelcontextprotocol/ext-apps.git /tmp/mcp-ext-apps
Read JSDoc documentation directly from `/tmp/mcp-ext-apps/src/`:
| File | Contents | |------|----------| | `src/app.ts` | `App` class, handlers (`ontoolinput`, `ontoolresult`, `onhostcontextchanged`, `onteardown`), lifecycle | | `src/server/index.ts` | `registerAppTool`, `registerAppResource`, tool visibility options | | `src/spec.types.ts` | All type definitions: `McpUiHostContext`, CSS variable keys, display modes | | `src/styles.ts` | `applyDocumentTheme`, `applyHostStyleVariables`, `applyHostFonts` | | `src/react/useApp.tsx` | `useApp` hook for React apps | | `src/react/useHostStyles.ts` | `useHostStyles`, `useHostStyleVariables`, `useHostFonts` hooks |
Learn and adapt from `/tmp/mcp-ext-apps/examples/basic-server-{framework}/`:
| Template | Key Files | |----------|-----------| | `basic-server-vanillajs/` | `server.ts`, `src/mcp-app.ts`, `mcp-app.html` | | `basic-server-react/` | `server.ts`, `src/mcp-app.tsx` (uses `useApp` hook) | | `basic-server-vue/` | `server.ts`, `src/App.vue` | | `basic-server-svelte/` | `server.ts`, `src/App.svelte` | | `basic-server-preact/` | `server.ts`, `src/mcp-app.tsx` | | `basic-server-solid/` | `server.ts`, `src/mcp-app.tsx` |
| Example | Relevant Pattern | |---------|-----------------| | `examples/map-server/` | External API integration + CSP (`connectDomains`, `resourceDomains`) | | `examples/sheet-music-server/` | Library that loads external assets (soundfonts) | | `examples/pdf-server/` | Binary content handling + app-only helper tools |
Before writing any code, examine the existing web app to plan what needs to change.
1. **Data sources** — How does the app get its data? (URL params, API calls, props, hardcoded, localStorage) 2. **External dependencies** — CDN scripts, fonts, API endpoints, iframe embeds, WebSocket connections 3. **Build system** — Current bundler (Webpack, Vite, Rollup, none), framework (React, Vue, vanilla), entry points 4. **User interactions** — Does the app have inputs/forms that should map to tool parameters? 5. **Runtime detection** — How to tell if the app is running inside an MCP host (e.g., check the current origin, a query param, or whether `window.parent !== window`)
Present findings to the user and confirm the approach.
In hybrid mode, the app keeps its existing data sources for standalone use and adds MCP equivalents:
| Standalone data source | MCP App equivalent | |---|---| | URL query parameters | `ontoolinput` / `ontoolresult` `arguments` or `structuredContent` | | REST API calls | `app.callServerTool()` to server-side tools, or keep direct API calls with CSP `connectDomains` | | Props / component inputs | `ontoolinput` `arguments` | | localStorage / sessionStorage | Not available in sandboxed iframe — pass via `structuredContent` or server-side state | | WebSocket connections | Keep with CSP `connectDomains`, or convert to polling via app-only tools | | Hardcoded data | Move to tool `structuredContent` to make it dynamic |
MCP Apps HTML runs in a sandboxed iframe with no same-origin server. **Every** external origin must be declared in CSP — missing origins fail silently.
**Before writing any code**, build the app and investigate all origins it references:
1. Build the app using the existing build command 2. Search the resulting HTML, CSS, and JS for **every** origin (not just "external" origins — every network request will need CSP approval) 3. For each origin found, trace back to source:
4. Check for third-party libraries that may make their own requests (analytics, error tracking, etc.)
**Document your findings** as three lists, and note for each origin whether it's universal, dev-only, or prod-only:
If no origins are found, the app may not need custom CSP domains.
Create a new MCP server with tool and resource r
CLI Agent Orchestrator (CAO) coordinates multiple AI coding CLIs so a supervisor can delegate work to specialist agents in parallel or sequence. 📚 Documentation — guides, reference, and two interactive courses.
Repo: awslabs/cli-agent-orchestrator
Author live dashboard UI from an agent via the `emit_ui` MCP tool. Emit
Find and select the best installed CAO agent profile for a task before
Enable, operate, and extend CAO's MCP Apps surface — the host-rendered fleet dashboard visible inside MCP App hosts (Claude Desktop, ChatGPT, VS Code Copilot,…
Store, recall, and forget durable facts with CAO memory — user preferences,
Create a new CAO (CLI Agent Orchestrator) plugin. Use this skill whenever the user wants to add a plugin that reacts to CAO lifecycle or messaging events,…