Turbo Browser MCP uses Playwright to drive Chromium. MCP is the standard for connecting agents to tools; an agent is a model running a multi-step task with tools.
Its ten tools include batched form filling, parallel tabs and page snapshots that can return only changes since the previous inspection. It can also upload a local file by path.
The README says form filling stops before submission by default. Its connection options include attaching to an existing browser session or starting a clean one.
In the author’s local Chromium test, filling a ten-field form and returning a snapshot took one tool call and 811 tokens, versus 11 turns and 12,197 direct tokens in the chosen step-by-step comparison.
Caveat: The performance figures are the author’s own measurements on a specific test, and a submit guard is not a substitute for reviewing an assistant’s browser permissions.
(HN 38 points · 7 comments · project README)
Lovable application exposure study — STALE: the updated analysis has no clear update date, and yesterday’s edition already cut the same report; a new HN discussion alone does not establish a new finding.
Language and AI energy study — STALE: the university’s account is dated November 5 and was also cut yesterday; no newer finding was verified.
Using GitHub Copilot with a local model — UNVERIFIABLE: the author’s Bluesky post links the article, but the article blocked extraction, so its setup and privacy details could not be checked.
A proposed EU model — UNVERIFIABLE: the Reddit headline alone does not establish the project details, and the discussion could not be extracted.