A single-file food delivery demo showcasing WebMCP, the proposed web standard that exposes structured tools for AI browser agents.
Repo: https://github.com/Doriandarko/webmcp-starter
9 WebMCP tools covering a full DoorDash-style ordering flow:
| Tool | API | What it does |
|---|---|---|
search_restaurants |
Imperative | Filter by cuisine, name, dietary tag |
get_menu |
Imperative | Browse a restaurant's full menu |
add_to_cart |
Imperative | Add items with quantity + special instructions |
remove_from_cart |
Imperative | Remove items from cart |
get_cart |
Imperative | Read-only cart state + totals |
clear_cart |
Imperative | Empty the cart |
apply_promo_code |
Imperative | Apply WELCOME20, FREEDELIVERY, or SAVE5 |
get_order_status |
Imperative | Track a placed order |
checkout |
Declarative | HTML form with address, tip, payment |
- Install C146+ (Canary)
- Enable
chrome://flags/#enable-webmcp-testing - Install the Model Context Tool Inspector extension
- Open
food-app.htmllocally in Chrome Canary - Click the extension → see all 9 tools → test with Gemini
Paste this into the extension's agent prompt:
I'm craving Italian food. Find me an Italian restaurant, add a margherita pizza, a carbonara, and a tiramisu. Apply WELCOME20. Checkout to 742 Evergreen Terrace, Apt 3B with a $5 tip on Visa.