Show HN: ApolloOffice – Local AI workers that turn GitHub issues into tested PRs

A developer released ApolloOffice, an open-source tool that runs AI agents in a visual office to pick up GitHub issues, reproduce them, and submit tested pull requests, available as a service on macOS, Linux and Windows. The developer said the tool produced 361 total pull requests on the ArtCraft project, 234 of which were merged, and that about 60% of pulls are code problems the tool found on its own. On low usage settings it generates roughly 51 pull requests a day, splitting tasks across Haiku models rather than a single Opus session to cut token usage.

This is a little thing I built: it's a visually pleasing office with AI agents working on GitHub repositories and submitting pull requests. They are getting issues, reproduce them and then fix them. I was using it privately in a Discord call and many people requested a public open-source version, so I got Claude Opus 5.5 to rewrite the entire code in Rust. It's available as a service on macOS, Linux and Windows. In my use case I use it to make PRs for ArtCraft, the Adobe Suite clone that is viral right now. This tool got me into the top 10 contributors within 5 days. I have 361 total pull requests, 234 of them got merged. I think that statistic speaks for the quality this thing can achieve. Right now on low usage settings it's round about 51 pulls a day fast, and it does not use up as much usage as you think it would, mainly because the tasks get split between Haiku models on default settings and not everything is running on one big Opus session. Also they get optimal prompts and clear instructions every step, which further reduces token usage. It can also bughunt: about 60% of pulls are code problems the tool found by itself. If you are a GitHub contributor to any project or want to solve issues on your own repository, this is the tool for you. Comments URL: https://news.ycombinator.com/item?id=50152285 https://news.ycombinator.com/item?id=50152285 Points: 1 Comments: 0