# A curated list of WebMCP demos, libraries, and tools

> Source: <https://github.com/GoogleChromeLabs/webmcp-tools/blob/main/AWESOME_WEBMCP.md>
> Published: 2026-09-01 20:03:18+00:00

A curated list of awesome WebMCP demos, libraries, and tools.

[Explainer mini-site](https://googlechromelabs.github.io/webmcp-tools/demos/explainer/)- A side-by-side demo of how AI agents interact with a web page today (site scraping) and a future where the page declares WebMCP structured tools.[Le Petit Bistro](https://googlechromelabs.github.io/webmcp-tools/demos/french-bistro/)- A restaurant reservation system that demonstrates how an AI agent can interact with a web-based booking form using declarative tool definitions.**Example Prompt:**"Can you book a table for John Doe for 2 people at Le Petit Bistro next Friday, at 7 PM on the Terrace? His number is 123-456-7890. It's for a birthday celebration."

[React Flight Search](https://googlechromelabs.github.io/webmcp-tools/demos/react-flightsearch/)- A React-based flight search application that showcases how a web application can expose structured tools to an AI agent, allowing it to programmatically interact with the UI.**Example Prompt:**"Search flights from LON to NYC leaving next Monday and returning after a week for 2 passengers."

[WebMCP zaMaker!](https://googlechromelabs.github.io/webmcp-tools/demos/pizza-maker/)- An interactive pizza builder demonstrating imperative tool registration, allowing AI models to make custom pizzas.**Example Prompt:**"Make me a large BBQ pizza with sauce, pineapple and extra bacon."

[Mystery Doors](https://googlechromelabs.github.io/webmcp-tools/demos/doors/)- A simple multi-page demo showcasing WebMCP capabilities using both declarative and imperative approaches.**Example Prompt:**"Explore the rooms"

[WebMCP Maze](https://googlechromelabs.github.io/webmcp-tools/demos/webmcp-maze/)- A maze escape game where the player navigates entirely by prompting an AI agent in the browser — no keyboard or mouse input.**Example Prompts:**"Start a new game", then "Solve the maze"

[CineFlow](https://googlechromelabs.github.io/webmcp-tools/demos/ticket-booking/)- A movie ticket purchase flow where AI agents can update location, query movies by genre, and select showtimes to start a checkout process.**Example Prompt:**"Find two tickets for a horror movie in Montpelier tonight"

[Order Tracking](https://googlechromelabs.github.io/webmcp-tools/demos/order-tracking/)- A simulated e-commerce order tracking and returns system allowing an AI agent to query order history and initiate returns via declarative tools.**Example Prompt:**"Where is my order from last week? If it’s delivered, I need to start a return."

[Open for Agents Storefront](https://demo.openforagents.com/)- A live WordPress and WooCommerce demonstration of owner-reviewed product-discovery tools, alongside an optional site Assistant that uses reviewed capabilities.[WordPress.org plugin](https://wordpress.org/plugins/open-for-agents-ai-toolkit-with-mcp/)|[Project explanation](https://www.openforagents.com/why-agent-ready)**Example Prompt:**"Find an in-stock mug."

[L'Atelier Hotel Chain](https://googlechromelabs.github.io/webmcp-tools/demos/hotel-chain/)- A high-fidelity hotel booking application designed to showcase the power of WebMCP. This demo illustrates how AI agents can interact with a modern web application through both imperative and declarative tools.**Example Prompt:**"Find a hotel in Paris with a Spa."

[WebMCP Sports](https://googlechromelabs.github.io/webmcp-tools/demos/sport-shop-angular/)- A modern e-commerce storefront for high-performance sports equipment, built with Angular and Vanilla CSS.**Example Prompt:**"Find training balls and add them to cart."

[Shoe store](https://andreinwald.github.io/webmcp-demo/)|[Code](https://github.com/andreinwald/webmcp-demo)- online store built with React, with all core functionality MCP-ready.**Example Prompt:**"Suggest the 3 best pairs of soccer shoes (foot size 45) available on this site. Add (one of suggestions) to cart and complete purchase."

[Flight booking](https://webmcp-flight-demo.netlify.app/)/[Flight booking (declarative)](https://webmcp-flight-demo.netlify.app/declarative.html)- Find the best flights for your journey.**Example Prompt:**"I want to book a flight from New York to Los Angeles for two people on next Thursday."

[Animal Viewer](https://65s6dw.csb.app/)- A simple codesandbox demo page that shows either a dog or a cat image.**Example Prompt:**"Show me a dog on this page"

[React Chess](https://matipojo.github.io/WebMCP-React-Chess)- A chess game that exposes WebMCP tools (`get-board-state`

,`make-move`

,`get-possible-moves`

,`restart-game`

,`promote-pawn`

) so an AI agent can play chess through`document.modelContext`

.**Example Prompt:**"Let's play chess. You play white. Make your opening move."

**Moving Beyond Screen Scraping**: A hands-on example of using WebMCP to create an agentic first experience with 10x fewer tokens[AI Audit](https://audit.wordlift.io/)- Audit any website's readiness for the agentic web. Exposes a`run-audit`

tool via`document.modelContext`

that lets an AI agent programmatically trigger a full AI readiness analysis (score 0–100) covering site files, SEO, structured data, content, and more.**Example Prompt:**"Run an AI readiness audit on[https://example.com](https://example.com)"

[Blackjack Agents](https://webmcp-blackjack.heejae.dev/)- Blackjack game with multiple AI agents (player, opponent, dealer) all using WebMCP tools. Each agent autonomously observes its hand, decides to hit or stand, and repeats until done — driven by tool descriptions alone.[Code](https://github.com/happyhj/webmcp-blackjack)**Example Prompt:**"Play my turn"

[WebMCP Bridge](https://h3manth.com/ai/webmcp/)- A bridge that connects any remote MCP server to Chrome's WebMCP API (`document.modelContext`

), letting browser-based AI agents discover and invoke tools from existing MCP servers.**Example Prompt:**"Search for TC39 proposals related to decorators"

[WebMCP × Excalidraw x WebAI](https://shidh.in/demo/webmcp-excalidraw/)- A Web app that converts natural language descriptions into Excalidraw diagrams through a 3-tool WebMCP pipeline (generate_mermaid → validate_mermaid → render_excalidraw), with optional on-device generation using Chrome's built-in AI.**Example Prompt**: "Create a flowchart showing the user login flow with error handling"

[WebMCP Flow](https://webmcp-flow.vercel.app/)- An AI-controllable architecture diagram builder that lets an AI agent create nodes, connect them with edges, and apply auto-layout in real time via WebMCP tools.[Code](https://github.com/ttimur-dev/webmcp-flow)**Example Prompt:**"Draw a typical web application architecture with authentication: browser client, API Gateway, Auth Service, User Service, PostgreSQL, Redis. Connect them with edges labeled by protocol and apply auto layout."

[The Morning Ritual (Imperative)](https://googlechromelabs.github.io/webmcp-tools/demos/coffee-shop/)- A premium coffee boutique demo showcasing agentic reordering, technical spec analysis, and catalog navigation.**Example Prompt:**"Will the espresso machine fit under a 15-inch cabinet?"

[UrbanEstates](https://googlechromelabs.github.io/webmcp-tools/demos/real-estate-map/)- A real-estate map application that demonstrates imperative tool registration, allowing an AI agent to interact with property filters and map views.**Example Prompt:**"Find me an apartment in Austin", then "Find me an apartment with AC under $1,000,000."

[Luxe Leather](https://googlechromelabs.github.io/webmcp-tools/demos/leather-bag/)- A premium e-commerce site for hand-crafted leather goods built with Angular and WebMCP. This demo showcases how an AI agent can interact with a web application to search products, check policies, and manage a shopping cart using declarative tools.**Example Prompt:**"Find a handmade leather bag, check if they have a 30-day return policy, and if yes, add the brown one to my cart"

[WebMCP Smart Home](https://googlechromelabs.github.io/webmcp-tools/demos/smart-home/)- A sleek, interactive dashboard where an AI agent can dynamically reconfigure home control components (e.g., front door cameras, thermostats, smart lights, energy distribution) based on user intents.**Example Prompt:**"Someone is at the door. Show me."

[webmcp.cool](https://webmcp.cool/)- A live, curated directory of WebMCP-enabled websites with a JSON API for agent-side discovery. Also registers its own WebMCP tools so agents can interact with the registry directly.**Example Prompt:**"List my site`https://example.com`

in the WebMCP directory."

[WebMCP Page Agent](https://googlechromelabs.github.io/webmcp-tools/demos/page-agent/)- A Gemini-powered meta-demo that lets you control any WebMCP-enabled website using simple natural language commands.**Example Prompt:**"Make me a large BBQ pizza with sauce, pineapple and extra bacon."

[JSON-stat WebMCP Explorer](https://jsonstat.com/webmcp/)- A JSON-stat viewer that lets you fetch a dataset from an official statistical office (like Eurostat), view and filter its data as a list or a cross-tabulation and download them as CSV using a web interface or natural language.**Example Prompt:**"Load dataset[https://ec.europa.eu/eurostat/api/dissemination/statistics/1.0/data/lfsi_emp_a?lang=en&lastTimePeriod=3&indic_em=ACT&age=Y15-64&unit=THS_PER](https://ec.europa.eu/eurostat/api/dissemination/statistics/1.0/data/lfsi_emp_a?lang=en&lastTimePeriod=3&indic_em=ACT&age=Y15-64&unit=THS_PER)and select the data for Germany and Ireland and the last period available. Display a cross-tabulation with sex as rows and countries as columns and download a CSV file with this info."

[Stacktree](https://stacktr.ee/)- Private hosting for the HTML that AI agents generate. Registers`stacktree_publish_html`

via`document.modelContext`

, so a browser agent can hand over a complete HTML document and get back a live, shareable URL with no account or sign-in. Anonymous sites are unlisted and the response includes a claim URL to keep the page permanently in a free account.**Example Prompt:**"Build a simple one-page site for a coffee shop called Crema and publish it so I can share the link."

(**Wordup**[live](https://snugug.github.io/demos/wordup/)): A casual word guessing game using the Prompt API to generate words and can be driven entirely through WebMCP.**Example Prompt:**"Start a new game."

[Latch](https://github.com/r0bertini/latch)- A one-line`<script>`

that detects a page's existing search, cart, and form handlers and registers them as WebMCP tools, with feature detection and no framework dependency. MIT licensed; running live at[latch.tools](https://latch.tools).[use-webmcp-tool](https://www.npmjs.com/package/use-webmcp-tool)- React hook for registering WebMCP tools (`document.modelContext`

) with lifecycle-managed registration.[webmcp-types](https://www.npmjs.com/package/webmcp-types)- TypeScript type definitions for WebMCP.[WebMCP - Model Context Tool Inspector](https://github.com/beaufortfrancois/model-context-tool-inspector)- A Chrome Extension to let web developers inspect web pages to verify if WebMCP tools are correctly exposed, visualize the input schema, and debug connection issues directly within the browser.[webmcpify](https://github.com/TueJon/webmcpify)- An agent skill that integrates WebMCP into an existing web app end to end: it inventories the app, proposes a tool manifest for approval, integrates the tools, then verifies each one in a real browser and heals failures.[MCP Webcomic Site Server](https://github.com/nearestnabors/mcp-webcomic-site-server)- A template and tutorial for making a webcomic archive visible to AI agents across three surfaces: a static 11ty website, an MCP server, and WebMCP browser tools. Registers tools like`get_current_page`

,`get_transcript`

,`prev_page`

/`next_page`

, and`search_comics`

via`navigator.modelContext.registerTool()`

.[WebMCP Kit](https://github.com/nekuda-ai/webmcp-kit)- A plugin for coding agents with an interactive visual Explorer that maps a site's user journeys to proposed WebMCP tools for review and approval, then implements and verifies them in a real browser.[WindTunnel](https://github.com/nekuda-ai/WindTunnel)- An open-source benchmark comparing WebMCP with other browser-agent interfaces across task success, execution time, token usage, and cost.

Contributions are welcome! Please read the [contribution guidelines](/GoogleChromeLabs/webmcp-tools/blob/main/CONTRIBUTING.md) first.
