@sarah_edo @sebastienlorber As a first step, here’s the polyfill via the WebAI Extension:
Stay tuned for a DevTrial in Chrome Canary!
Let's talk about good WebMCP tool descriptions! We'll write these like a good commit message, not a README. We don't want it long and polluting context windows. Let's say what it does and when to use it
Positive framing: "Searches products by keyword."
Not: "Do not use this for orders."
The model will read each word, let's make them count ☺️
Show more
jev + opus 5.5... i simply can't comprehend why everyone isn't building this yet.
in my workflow, this cut costs and time by ~80%. i think it's one of the best ways to use it.
→ pick relevant project notes before loading the context
→ route suitable tasks to a faster worker
→ choose a recovery path when a tool fails
→ run focused checks before the full test suite
opus handles the hard reasoning. jev picks from options the harness prepares and validates.
i explain how to build the decision layer in the article below:
Show more
AI is getting cheap enough that you can process everything you see
Built a Thrift Store Scanning Agent that processes video frames, looks up the price online and alerts you to good deals
Show more
🔄 Every software loop needs feedback: design, build, test, launch, and review.
When human users hit broken flows on a site, some file support tickets. But when AI agents hit dead ends, they fail silently or apologize in prose.
What if we give agents a bug-reporting tool? 🧵👇
Show more
Hi internet! We went ahead and upgraded all @stripe hosted checkout pages for 7.8M businesses—0.45% of the world’s GDP—with WebMCP so agents can more efficiently buy. Our evals show this reduces checkout latency by 39% and token usage by 42%.
Learn more:
Show more
🎇 I made a small Chrome extension: a side panel that drives any site's WebMCP tools with Jev!
When you type, on every keystroke it picks the relevant page's tool, fills in the arguments, and tells you how sure it is.
Here it is grocery shopping:
Show more
I’ve been experimenting with Jev and WebMCP 😁 - Jev selects the next tool and fills enumerable inputs. If the same call needs free text or a number, it falls back to LLM.
If that same call also needs free text, a number, or another non-enumerable value, an LLM fills only those remaining fields.
What's also interesting is that based on the confidence score given by Jev we can make decisions in harness - if Jev is 97% sure we might trigger the tool - but if the confidence is, let's say, 60% we might want to do additional checks!
Evals next? 🤔
1/2 🧵
Show more
@sarah_edo Showing the confidence per tool is the clever bit. It turns the agent from a black box into something you can actually debug, and it tells you when your tool descriptions are stepping on each other.
Show more
This makes Jev click for me now. It’s not really about replacing frontier models, it’s about not needing one for every decision an agent makes. Let Jev handle the fast, repetitive decisions and bring in the frontier model when deeper reasoning or generation is actually needed. Maybe the agent stack ends up being less about one model doing everything and more about using the right intelligence for the right decision.
Show more
🎇 I made a small Chrome extension: a side panel that drives any site's WebMCP tools with Jev!
When you type, on every keystroke it picks the relevant page's tool, fills in the arguments, and tells you how sure it is.
Here it is grocery shopping:
Show more
How do you secure an ever-evolving open source codebase with 2,300+ third-party dependencies in the AI era?
@googlechrome security leads @dougturner & Jasika Bawa share some insider details on what we're doing & our multi-pronged strategy to make Chrome and the web safer in 🕸️✨💪
Show more
🔥 It was an absolute pleasure to talk about WebMCP among other AI on the web topics with @domfarolino on the latest @syntaxfm!
Along with the rest of the internet, @wesbos and @stolinski are heroes of mine. Thank you for hosting us!
Show more
I'll be at this office hours this morning!
The WebMCP Challenge is here.
We’ve teamed up with @ChromiumDev, @CloudflareDev, @ShopifyDevs, @vercel, @render, and @Netlify for a 10-day hackathon.
Up for grabs: $35,000 in cash prizes, Codex Micros, ChatGPT Pro subscriptions, and more prizes from our supporters.
Show more
Create agent-ready web apps for the WebMCP Challenge →
We're excited to see ChatGPT supporting WebMCP, the experimental standard that lets users and AI agents navigate websites together. In the @OpenAIDevs WebMCP Challenge you can experiment with it and win prizes. Plus, Chrome's @sarah_edo is on the judging panel! What will you create?
Show more
📣 WebMCP Hackathon!
If you build an innovative experience leveraging WebMCP, you can win prizes! People from across the industry are coming together for a virtual hackathon.
I’m excited to participate as one of the judges, I’ll look out for game-changing implementations.
1/2
Show more
It would be easier to use caffinate if I knew how to spell caffeinate
Yo mutuals, I am looking for a designer to help design materials for my sister's funeral. I'm happy to pay if necessary, but please reach out to me in DMs please. Need a quick turnaround as well (hopefully by Monday). Thank you!
Show more
Do not sell the whole ticket at once. @shaundai on micro-sales: get the conversation, then the next yes.
