Vinicius Dallacqua

Vinicius Dallacqua

@webx.bsky.social

Into browsers and agentic systems. Engineering QA.tech Building https://agent.perflab.io Into film photography and web performance nerd.

970 followers205 following202 posts

Pinned

Hello everyone, I'd like to give you all a quick update on the PerfAgent progress! It's been a wild ride since the inception of PerfLab, from forking DevTools to build a new experience using all of its power and trying to take my owns spin on Performance Tooling for all devs! 🧵

Vinicius Dallacqua reposted

It's happening... We've asked Chromium API owners for permission to ship The Soft Navigation work to allow measuring of Core Web Vitals for SPAs!!

WebMCP is a great new spec and I've been playing around with it in my own implementation into my web browsing agent harness. It brings interactions to a fraction of the time and sessions get faster and also 'cheaper' for the models.

Vinicius Dallacqua reposted

This React -> Web Components post by the MDN team touches on challenges and solutions I've seen dozens of times over frontend's lost decade. Kudos to them for the improvements in performance, and the bravery to tell the story in an unvarnished way: developer.mozilla.org/en-US/blog/m...

developer.mozilla.orgUnder the hood of MDN's new frontend | MDN BlogYou may have spotted that MDN has a new frontend. There's plenty happening under the surface, so let's unpack the technologies we chose, the architectural decisions we made, and why we did a rebuild…

I've been playing with building a personal assistant meets browsing assistant for mobile web and had to experiment with WebMCP as part of the experience. It's amazing to see the experience when the browsing agent performs tasks instantly instead of wasting turns to perform a task.

Tthe baseline for the browsing agent I've built can navigate around and perform tasks for me but takes several turns, spends more tokens and takes more time. After implementing WebMCP into the harness, tasks that took several turns became instant (for pages that exposes the features)

Vinicius Dallacqua reposted

myth: LLM-assisted coding means everyone is forced to use The Popular Framework forever reality: agents will happily migrate your entire app to The Good Framework inertia is a powerful drug but we are entering a world without moats, plan accordingly strawberrybrowser.com/blog/react-t...

strawberrybrowser.comHow we Rewrote 130K Lines from React to Svelte in Two WeeksWe ported Strawberry's entire frontend from React to Svelte in two weeks using coding agents. It made the browser 2x faster.

Vinicius Dallacqua reposted

When a Skill activates, its SKILL.md loads into context. The webperf-loading was 24 KB, most of it unnecessary to decide which script to run I applied progressive disclosure and brought it down to 11 KB #WebPerf #Performance #AI #AIAgent #AIPerf

I have a 'neuve' side project and I'll start writting more about it here. This time the writeup brings some of my experience when developing the browsing aspect of, I stopped telling AI agents what to build.

View full thread

They converged on similar patterns I've seen other browsing agents use under the hood. But some of the findings and the process I took taught me quite a lot. On the YT link you can see an early experiment of this browsing agent with Kernel.sh as the CDP session provider youtu.be/yzzec2KFI7c

kernel.shKernel: Browser infrastructure for web agents and automationsBrowser infrastructure for AI agents and web automations. Anti-bot detection, reusable sessions, fast spin-up, and autoscaling browsers.

Using Opus 4.5 I turned Claude Code and agent-browser into my personal AI lab and triggered differernt sub-agents and 'co-workers' to independently explore in a sandbox the best way to converge into a good set of guardrails, tools and CoT prompt autonomously Full article here dev.to/viniciusdall...

dev.toAgents That Build Agents — Building Autonomous Browsing with Claude CodeAgents That Build and Improve Agents — Building Autonomous Browsing with Claude...

I have a 'neuve' side project and I'll start writting more about it here. This time the writeup brings some of my experience when developing the browsing aspect of, I stopped telling AI agents what to build.

Instead, I carefully created an experiment on the problem. No shared context, no instructions and no specific method in mind and let them figure out how to drive a browsing session based on a goal independently.

I've actually used the learnings on how to compress trace information into a more token efficient format in a custom pipeline using Firecrawls's response (MD representation of a web page) into something more efficient, I'm calling CMD (Compressed Markdown DSL), but that's a writeup for another time.

Vinicius Dallacqua reposted

Here's your friendly reminder that CSS selectors are global and evaluated RTL, so if you have selectors that are very broad like `.container > div` you might be impacting performance in a measurable way 🙈 Before and after profile of server switching after some selector optimizations in @discord.com

Wrapping up the year with my article for the Perf Calendar with my write up about how I ended up creating agent.perflab.io! It has AI, it has Perf…what else could you want 😎🔥 I’m always happy to see my content in the Performance Calendar! Thanks for the initiative @stoyan.me !

agent.perflab.ioPerfAgent - AI-Powered Web Performance AnalysisGet expert insights on Core Web Vitals and performance optimization from an AI specialized in web performance analysis.

Web Performance Calendar day 31 article 3/5: Can AI meaningfully help with real-world performance analysis? @webx.bsky.social explores how to train AI/LLM agents to interpret complex browser data (like DevTools traces) into actionable insights. calendar.perfplanet.com/2025/teachin...

Vinicius Dallacqua reposted

Quite the journey!

Chrome for Developers@developer.chrome.com

Proactively speed up future requests with dns-prefetch → goo.gle/4q1uO3I Now part of Baseline, this hint tells the browser to perform DNS lookups for other origins in the background, reducing latency for your users.

I’m super hyped and happy to announce that I’m getting even deeper into helping agents reason about software quality as I join QA.tech! Since I’ve started working on PerfLab and then PerfAgent this has become an obsession. Now I get to do that full-time and I could not be more thrilled!

Vinicius Dallacqua reposted

something very cool to see: Chrome recently shipped automatic <link rel="prefetch"> for all links in viewport you could do it previously with libs like Quicklink, but it’s super cool to see this implemented natively (it’s off by default, enabled with one html tag)

Vinicius Dallacqua reposted

✨SPDY STREAM 010 ✨ with Jacob Groß 📢 As mentioned during yesterday's stream, Jacob is our next guest, and we shall cover tag mgmt and the deleterious effect they can have on performance, esp web vitals and possible solutions. Join us! 📆 Wednesday Sept 17th ⏰ 1PM EDT 🔗 bit.ly/spdy-stream-010

bit.lySPDY STREAM with Jacob Groß 🇩🇪In this episode, we are joined by Jacob Groß to discuss tag managers and their often detrimental effect on website performance. We will explore how tag managers can impact performance in a multitude o...

Hello everyone, I have just heard that the company I'm a lead engineer for is going to downsize so I'm going to look for the next thing from next week. I'm looking to work with AI engineering and am shipping lots of different code automations and agent workflows.

Am currently trying to get my own thing off the ground prototyping an intersection of browsers and personal assistants and also autonomous coding agents that are capable of self improving. Hit me up know if you know of any oportunities around 🫡

I'm also open for consuntancy, so reach out if you are looking to collaborate. I've been building v0.perflab.io agent.perflab.io and other cool stuff, with V0 SDK, AISDK and Mastra Also have interest on browsers and DevTools with my own projects and forks of it to ship my own autonomous agents.

v0.perflab.ioV0 MCP ServerModel Context Protocol (MCP) server that exposes v0.dev APIs through the Streamable HTTP pattern. Built with Hono framework and deployed on Vercel.

I'm beyond excited to share a stage with so many great names! This is a special event to me and I'm super hyped to share a bit of my story with you all ❤️

performance.now() is coming! @webx.bsky.social [] will explain how he built a Performance AI assistant and agentic workflow on top of a fork of Chrome DevTools. Few AI systems can reason with the complex reality of performance data. Learn how to solve that. perfnow.nl/speakers.htm...

Vinicius Dallacqua reposted

Really excited to be speaking @perfnow.nl this year—my favourite conference run by some of my favourite people! If you're into web performance (or wanna be!) and like a great conference with a super friendly vibe then pop along to Amsterdam at the end of October. Tickets selling fast! perfnow.nl

perfnow.nlperformance.now(); 30th and 31st of October, AmsterdamOn 30th and 31st of October, 2025, the sixth edition of performance.now() will take place in Amsterdam—a single track conference, covering today’s most important web performance insights.

Whist trying out my demo for the Agentic Perf Analysis and Debugging, I managed to use my V0 and my DevTools MCP servers to debug one of my V0 demo chats. Using animation frame insights no less 😃

Vinicius Dallacqua reposted

performance.now() is coming to Amsterdam this October! This event is great both for the inspirational talks as well as for the opportunity to meet folks from different companies and backgrounds to discuss how to successfully tackle web performance challenges!

performance.now(), the web performance conference, 30th and 31st of October, Amsterdam. Single track, the best speakers in the industry, plenty of your performance-mad peers around to have fascinating conversations. What's not to like? Let's introduce our speakers ... perfnow.nl

Vinicius Dallacqua reposted

performance.now(), the web performance conference, 30th and 31st of October, Amsterdam. Single track, the best speakers in the industry, plenty of your performance-mad peers around to have fascinating conversations. What's not to like? https://perfnow.nl/

perfnow.nlperformance.now(); 30th and 31st of October, AmsterdamOn 30th and 31st of October, 2025, the sixth edition of performance.now() will take place in Amsterdam—a single track conference, covering today’s most important web performance insights.

We're almost there! After spending the week hacking OAuth support into Mastra MCP client class and getting the toolcall aproval workflow to an MVP state this sprint is soon coming to an end.

Vinicius Dallacqua@webx.bsky.social

Updates on PerfAgent, after spending some time implementing v0.perflab.io MCP server it's now time to create PerfAgent's MCP client. It will be a great way to expose the agent to external services and agents such as @sentry.io Seer, and my own servers too, like V0.

There's a larger goal of mine here, my experiments with MCP is to try to better understand how to enable good primitives for AI native systems.

Vinicius Dallacqua@webx.bsky.social

Updates on PerfAgent, after spending some time implementing v0.perflab.io MCP server it's now time to create PerfAgent's MCP client. It will be a great way to expose the agent to external services and agents such as @sentry.io Seer, and my own servers too, like V0.

View full thread

I've managed to implement OAuth into Mastra MCP client (on a fork of mine) and now it talks to my V0 remote MCP server (that uses OAuth). Gotta work on the HITL parts of the MCP tool call as the spec requests, but we're moving now 😎.

I've started to implement the client using Mastra MCP client, since I already use Mastra as my agent framework with AI SDK. But Mastra supports only setting the auth header, and do not (at this time) support the full OAuth flow. Good thing is, I've been playing lots with OAuth flow for MCP servers