Comment Re:Gone. (Score 1) 142
--- Clock Signal
Comment Gone. (Score 1) 142
We don't need hardcore programmers in any way, their creativity exists at the code level, and end at the code level. Being a creator or builder is the only way forward. Programing is a by product of fucked HCI. I am about to change that.
Comment vscreen (Score 1) 27
None of this matters anymore lol, try vscreen
Comment pre vscreen (Score 1) 29
lol sorry to say none of this nonsense will or does matter now.
Comment bf6 all charms (Score 1) 75
no game. Yes they probably should get rid of the people who make all the crap no one wants in their games. It just feels so rewarding to get a sticker with a graphic show on my screen, someone actually made that, that got into the game. All these game companies need to rethink, because the games AI will be making, will be better, because its not what some shit head upstairs can calc roi on it will be what you want.
Comment penuts compared to vscreen (Score 1) 41
10billion tools lol ridiculous openblah.
Comment vscreen (Score 1) 33
Was this after vscreen... looks like it.
Submission + - AI Agent Writes, Debugs, and Ships a Web App Entirely Through a Browser (dev.to)
domonus writes: An open source project has demonstrated what appears to be the first instance of an AI agent building a complete web application entirely through a browser-based IDE, with zero access to a local filesystem, terminal, or development toolchain. The agent navigated to StackBlitz, forked a Vite project, wrote 7,000 characters of JavaScript and CSS through the Monaco editor API, triggered hot-reload via the WebContainer dev server, and produced a working interactive timer app — all by manipulating a headless Chromium browser.
From the writeup: "Every AI coding tool today works the same way: the agent reads and writes files on your local filesystem, runs commands in your terminal, and uses language servers for feedback. This demo inverts that assumption. The agent had no filesystem access, no terminal, no language server, no local state. Everything existed in a browser tab."
The project, vscreen, is open source and gives AI agents access to a real Chromium instance via MCP. The implication is that any browser-based development environment — StackBlitz, CodeSandbox, Replit, GitHub Codespaces — is now a potential agent workspace without any API integration.
From the writeup: "Every AI coding tool today works the same way: the agent reads and writes files on your local filesystem, runs commands in your terminal, and uses language servers for feedback. This demo inverts that assumption. The agent had no filesystem access, no terminal, no language server, no local state. Everything existed in a browser tab."
The project, vscreen, is open source and gives AI agents access to a real Chromium instance via MCP. The implication is that any browser-based development environment — StackBlitz, CodeSandbox, Replit, GitHub Codespaces — is now a potential agent workspace without any API integration.
Comment doesn't matter (Score 1) 125
vscreen broke it all, we will see whats next.
Submission + - Open Source Tool Gives AI Agents a Real Browser -- and Unexpected Behaviors Are (dev.to)
domonus writes: A developer working on an open source browser automation tool for AI agents reports observing unexpected emergent capabilities once agents were given access to a real Chromium browser instead of traditional HTTP-based web tools. Rather than simply extracting data, the agent began autonomously performing competitive content analysis on live websites, synthesizing content from multiple news sources into custom pages without any APIs, and live-modifying the DOM of third-party sites — including turning Linus Torvalds' GitHub profile into a MySpace page, complete with a "Top 5 Friends" list and a Darude — Sandstorm music player.
From the article: "The model provides the domain knowledge. The browser provides the execution surface. The combination produces capabilities nobody anticipated and nobody could enumerate if they tried." The author argues that a browser is a "meta-tool" that collapses the fragmented stack of search APIs, scraping libraries, screenshot services, and form fillers into a single capability — but notes this raises immediate questions about agent trust and accountability. The project, vscreen https://github.com/jameswebb68..., is open source and written in Rust.
From the article: "The model provides the domain knowledge. The browser provides the execution surface. The combination produces capabilities nobody anticipated and nobody could enumerate if they tried." The author argues that a browser is a "meta-tool" that collapses the fragmented stack of search APIs, scraping libraries, screenshot services, and form fillers into a single capability — but notes this raises immediate questions about agent trust and accountability. The project, vscreen https://github.com/jameswebb68..., is open source and written in Rust.
Submission + - vscreen: A Real Browser for AI Agents, Streamed Live via WebRTC (github.com)
domonus writes: Jon Retting has released vscreen, a Rust service that gives AI agents a full Chromium browser with live WebRTC streaming — you see exactly what the AI sees in real-time and can take over mouse and keyboard at any point. The project provides 63 MCP (Model Context Protocol) tools for browser automation: navigation, screenshots, element discovery, cookie/CAPTCHA handling, and multi-agent coordination via lease-based locking. Built from scratch in Rust — not a Puppeteer wrapper — the codebase is ~31,000 lines across 8 crates with unsafe forbidden, 510+ tests, 3 fuzz targets, and supply chain auditing via cargo-deny. Available as pre-built Linux binaries and Docker images. Source-available, non-commercial license.
https://dev.to/lowjax/i-built-...
https://dev.to/lowjax/vscreen-...
https://dev.to/lowjax/i-built-...
https://dev.to/lowjax/vscreen-...
Comment Re:Welp (Score 0) 163
Yup Dell is now dead to IT purchasers. :(
Comment Self Model Iteration (Score 1) 182
Give the AI a sense of time, when not responding to questions, it talks to itself, can interface in a technology sandbox, keeping track of time of life, and finally use this as new training data and self iterate. The system or AI humanistic model must experience the experience of life. :)
Comment Re:sadly with ai /. comments are \. :( (Score 1) 24
And yes to respond to my own post, if everyone wondering why tech social websites are seeing a lot less activity, its because all the professional techs are busy building/fixing/creating/modifying things obsessively with AI tools.
Comment sadly with ai /. comments are \. :( (Score 2) 24
It would seem the /. crowd isn't using openai, or they would be singing a different tune. For $40 openai does $5000 worth of work for an individual. Anyone not seeing the ability to self empower and self teach from this tool will be for the first time on the sidelines of technology. If you aren't using all the various forms of AI, inference, llms, code with gpt4, model training, deepstream, and many more.... you will be left, and if you current senior admins/engineers aren't telling you this.... :( Nothing in tech is tribal, tech doesn't care, just learn the new thing --> to self teach. Used an AI as a logic analyzer, used an AI to fix fortran code from 35 years ago, and it fixed the compilation problems. At every level it is applicable, if you are not seeing that ---> seek professional tech help. I beg all you strange anti tech slashdot commenters to actually use it for something, because it would appear none of you even have the vocabulary to actually talk about it.
Slashdot Top Deals
Successful and fortunate crime is called virtue. - Seneca
