Try it

WebMCP is an early preview, so setup takes a few minutes and suits people comfortable installing a browser extension. If you would rather wait, nothing is lost: it will get easier as browsers and assistants catch up.

Steps#

1. Use a recent Chrome. Chrome's origin trial starts from Chrome 149, and these steps work for local testing.
2. Turn WebMCP on. Open chrome://flags/#enable-webmcp-testing in the address bar, set the flag to Enabled, then relaunch Chrome.
3. Add an assistant that can use website tools. Chrome does not include one that is documented to call WebMCP tools yet, so testers use an extension. Chrome's documentation lists two:
   • The Model Context Tool Inspector from the Chrome Web Store. It shows the tools a page offers and lets you try them with prompts. It needs a Gemini API key.
   • The Example Agentic Chrome Extension on GitHub. It needs a small local server and your own API key, so it suits developers more than most students.
4. Open NTUMods and check the tools appear. Go to the timetable planner and open the extension. You should see about 32 NTUMods tools listed. If you see none, check that the flag is enabled and that you relaunched Chrome.
5. Ask for something specific, such as the examples, and check the result on the page.

New to this? Run the comparison demo with the same extension first. It is a small page with two tools, so it is an easy way to confirm your setup works before you try NTUMods.

Before you start#

• An API key is tied to your own account and usage. Only paste it into an extension you trust, and keep an eye on any usage limits.
• Gemini in Chrome is a separate feature, and Chrome's documentation does not say it can call WebMCP tools.
• You can read the full list of NTUMods tools with no setup in webmcp.json, or on the tools page.

What to read#

• WebMCP on Chrome for Developers: the overview, with setup and testing tools. Start here.
• Imperative API guide: how a site registers tools, which is the approach NTUMods uses.
• WebMCP on GitHub and the specification: the standard itself, where you can follow progress and report issues.