A live WebMCP playground

Give your browser hands.

This page doesn't hand an AI agent a screen to squint at. It hands it typed functions. Ask an agent to read this page, fill the form, fetch your local weather, make a share image, and post it. Watch every call arrive on the right.

checking…
READ · summarize_page()

Let the agent understand this page

The page hands over its own content as structured data, so nothing gets scraped off a rendered screen.

Nothing read yet. An agent would call summarize_page to learn what this page is and what it can do.
WRITE · fill_lead_form()

Request a walkthrough

The agent writes into the DOM with typed arguments. It can fill and propose, but a human still presses confirm. Nothing is sent anywhere.

FETCH · get_weather()

Pull in live context

A tool can reach a real API. This one calls Open-Meteo. With no arguments it uses your approximate location (from your connection, via the edge). The agent can also pass any city.

No data fetched yet. An agent would call get_weather.
GENERATE · create_share_card()

Make a shareable image

The agent renders a branded card on a <canvas> and produces a downloadable PNG: an artifact generated on the page.

No card generated yet. An agent would call create_share_card.
NAVIGATE · share_page()

Tell the world

A tool can trigger an outbound action, like opening the X or LinkedIn share composer. The human still chooses to post.

No post composed yet. An agent would call share_page.

Tool calls

idle
    Calls appear here the moment an agent (or you, via the DevTools
    console) invokes a tool. See “how to run it” up top.

    How to test WebMCP

    A page can publish tools, but something has to be the agent that calls them. Easiest path: install the tiny on-device agent below. It runs Chrome's built-in Gemini Nano, so there's no API key and nothing about you leaves your machine.

    Easiest · on-device · no key

    The WebMCP Agent extension

    A tiny Chrome extension that reads this page's tools and drives them with Gemini Nano, the model Chrome already ships. No API key, no network, no flags.

    ↓ Download the extension (.zip)
    1. Unzip it.
    2. Open chrome://extensions and turn on Developer mode (top-right).
    3. Click Load unpacked and choose the web-mcp-agent folder.
    4. Pin the extension, come back here, click it, type a goal, and press Run.
    Try:"Book a demo — I'm Priya at Acme, priya@acme.com, then make a share card and post it to X."

    Chrome 138+ on desktop. The first run downloads Gemini Nano once (a few GB); after that it runs instantly and offline. It's the client half of WebMCP, so it's an agent built on Chrome's model, not the browser doing it natively.

    With ChatGPT

    ChatGPT desktop app

    Its built-in browser calls site tools with GPT-5.6 Sol or Terra. Open this page as a tab and ask. Heads up: site tools are off on Enterprise/Edu workspaces, so use a personal account.

    Chrome 149+ / Edge 147

    Flip the flag

    Turn the API on, then drive it with the extension above or by hand.

    1. Open chrome://flags/#enable-webmcp-testing
    2. Set it to Enabled, relaunch
    3. Reload this page. The badge above turns green.
    Developers

    Call them in the console

    With WebMCP on, drive the tools yourself in DevTools: real calls, no agent.

    await document.modelContext.getTools()

    Read

    summarize_page returns data, not pixels. The page tells the agent what it is, so it reasons instead of scraping.

    Write

    fill_lead_form mutates the DOM with typed args, then arms the button. A human commits. The page chose that gate; the browser doesn't enforce it.

    Fetch

    get_weather hits a real external API. Tools aren't limited to the page. They can reach the internet.

    Generate

    create_share_card renders a canvas image and hands back a downloadable file, so an agent can produce a real artifact right on the page.

    Navigate

    share_page triggers an outbound action. The agent opens the composer; you decide to post.