Arena
Benchmark and compare the best AI models
Start new thread

Arena Agent Mode with GitHub - Get real work done, moving from idea to shipping in minutes

by
•
For a coding agent to be truly useful, it needs to live where the work is: on GitHub. By integrating directly with GitHub and reusing the underlying infrastructure of Agent Mode, we’ve created a seamless experience that allows you to connect a repo, complete a coding task, and push the results—all without leaving your browser.

Add a comment

Replies

Best
Maker
📌
👋 Hey Product Hunt! We're excited to launch Coding in Agent Mode 👨‍💻 AI coding assistants have fundamentally changed how code is created. But for most developers, a chat interface has been a "read-only" or "copy-paste" coding experience — until now! Today, we are excited to introduce Coding in Agent Mode, a web-based coding agent that transforms your browser into a full-featured, GitHub-connected development environment. This lets you get real work done, quickly going from idea to shipped feature in a matter of minutes.
Quick question. I cant find any pricing info etc. Is it free to use? Or how does it work?
Maker

 Yes, free to use!

 Then the followup question becomes, how will you make money later? whats the business model?

Maker

 Thanks for the question. If you scroll down to the bottom of our we have a section that talks about this, as well as links to our AI evaluations blog post to learn more

Not sure - what's the value proposition?

Maker

 Check out our launch blog here for more info!

 thanks, that clears up the value prop actually - a browser-based agent that can go end to end on a real GitHub-connected repo is a different thing than a chat window. one thing I'm curious about: when a session involves a long-running build or a dev server that needs to stay up while the agent iterates, is that running in a persistent cloud sandbox tied to the session, or does it spin up fresh each time and lose installed deps/running state if you step away and come back later?

Maker

 The sandbox is persistent and stays active throughout the agent run. When the agent finishes a turn, we pause the sandbox, which preserves all installed dependencies, files, and running state. It resumes when you continue chatting with the agent.

 good to know, that answers it. persistent sandbox makes a lot more sense for real dev workflows than a fresh container every message

Maker

You can learn more about this functionality in our blog post:

This is a solid step toward making agentic coding feel more like actually having a dev environment in your browser.