The network for creativity
Join 1.25M professional creatives like you
Connect with clients, get discovered, and run your business 100% commission-free
Creatives on Contra have earned over $150M and we are just getting started
Back to feedPost
Every MCP server you connect costs tokens before it does anything.
We measured five browser MCP servers reading the same page (one URL, 18 Sep):
• chrome-devtools-mcp: 12,321 tokens
• Playwright MCP: 11,119
• read-website-fast: 4,938
• server-puppeteer: 2,871
• mcp-server-fetch: 1,298
The tool list alone costs 5,760 tokens for DevTools and 4,366 for Playwright the moment you connect, before a single call.
Why it matters: those tool definitions sit in context on every turn. Connect three heavy servers "just in case" and you pay for all of them on every message, in cost and in the model's attention.
The rule we now use:
• Reading a page: fetch.
• Driving a page (clicks, forms, flows): Playwright.
• Debugging runtime (console, network): DevTools, connected only for that task.
If you're building agents for clients, this is an easy win to check first: list what's connected, and cut what the task doesn't need.
The network for creativity
Join 1.25M professional creatives like you
Connect with clients, get discovered, and run your business 100% commission-free
Creatives on Contra have earned over $150M and we are just getting started
Related posts
Behind the scenes on this one, because the build was more interesting than the result.
The task: get an AI agent to play a browser Tetris clone and beat the high score.
Problem one — the game renders to a <canvas>. No DOM, no list of filled cells, just pixels. So the first thing to build was a reader: divide the canvas into a 10×18 grid, sample each cell, decide filled or empty.
First attempt classified by colour, matching each cell against known block colours. Failed instantly and silently. Purple pieces read as empty. Gradient regions read as empty. The bot was playing against a board that was largely imaginary.
The fix was changing the question. Not "what colour is this cell" but "how much does this cell vary." Empty cells are flat. Filled cells have an edge, a bevel, a highlight. Measure luminance across the cell, take max minus min, threshold it. Works on every piece colour without enumerating any of them.
Then a flood fill to isolate the falling piece, and a placement search — for every column and rotation, simulate the drop, score the resulting board, pick the best.
Problem two — the browser protocol caps each execution at 45 seconds. Tetris doesn't finish in 45 seconds. So instead of an async loop that waits, you install an interval, return immediately, and poll the page every twenty seconds.
That's where it went wrong. The smart bot was too slow, so it got simplified down to a fixed key pattern. And the number being polled turned out to be the loop's own tick count, not the game score. Five runs of "improving" numbers that measured nothing at all.
The interesting part was the cleanup. To find out what actually happened, I went back to a 30-minute screen recording:
— ffmpeg to pull frames at 2fps
— locate the playfield in pixel coordinates
— classify every cell across a dense frame sequence
— then render the classifier's verdicts back over the frames and check them by eye
That last step is the one that mattered. The first pass flagged the bottom row as complete in almost every frame, which looked like a bug in my grid alignment. It wasn't. The row genuinely was complete, and the game genuinely never cleared it.
Verified: a completed line sits in the well for ten full seconds while the stack piles on top of it. Across ninety seconds of play the board empties about thirty times, and every one of those is a game over, not a line clear.
The lesson I'm taking into the next build: before you simplify under pressure, write down what you're optimising for, then check whether the simplification is still attached to it.
infiniteup.dev
The Tetris Clone Didn’t Clear Lines. Our AI Agent Didn’t Notice. - InfiniteUp
We pointed an AI agent at a browser game and it reported five runs of improving scores that measured nothing at all. It also reached a correct conclusion on worthless evidence. What that failure means for anyone deploying agentic AI.
tetris on a canvas is a tough one. did the agent beat the high score?
I built QuoteKit because quoting a project should not feel like guessing in public.
A client starts with a package, shapes the scope, chooses a timeline and explains what they need. QuoteKit turns those choices into a working estimate, a realistic range and clear reasons behind the price.
It works in English and French, adapts to mobile and gives both sides a summary they can copy, download or print.
I designed the product, pricing logic and interface for NexusHub Studio. It is a small tool with a simple goal: make the first client conversation clearer.
QuoteKit is available for 10 USD with a commercial license.
What would you want an estimator like this to explain before you send a quote?
Quote clearly. Start better.
Explaining the estimate turns pricing into a much clearer conversation.
Your website doesn’t have an attention problem.
It has an attention competition problem.
I recently saw a website with great typography, beautiful animations, strong case studies, and lots of useful information.
Nothing was wrong with it.
That was the problem.
Everything was competing for attention.
- The hero wanted attention
- Features wanted attention
- Animation wanted attention
- CTA wanted attention
So I started thinking about websites differently:
🚫 Don’t ask, “What should I add?”
✅ Ask, “What should win?”
That’s attention architecture.
Designing the hierarchy so the visitor knows what matters, what comes next, and what they can safely ignore.
A website doesn’t need to capture more attention.
It needs to waste less of it.
If your website feels busy but you can’t figure out why, book a free discovery call with me!
