ChatML - Run Claude Code Parallel Sessions in a Desktop app
Hey HN, Over the past 10 months I've been using Claude Code heavily, and one limitation kept coming up: you can really only run one coding agent at a time. While one agent is refactoring something, the rest of the repo is basically blocked unless you start manually juggling branches and working directories. The core issue is that AI coding agents operate directly in your filesystem. If two agents run in the same working directory they quickly start stepping on each other’s changes. Git worktrees turned out to be a surprisingly good primitive for solving this. So I built ChatML, a Desktop app…
What it does
In the maker’s words, at launch
Hey HN, Over the past 10 months I've been using Claude Code heavily, and one limitation kept coming up: you can really only run one coding agent at a time. While one agent is refactoring something, the rest of the repo is basically blocked unless you start manually juggling branches and working directories. The core issue is that AI coding agents operate directly in your filesystem. If two agents run in the same working directory they quickly start stepping on each other’s changes. Git worktrees turned out to be a surprisingly good primitive for solving this. So I built ChatML, a Desktop app that runs multiple Claude Code sessions in parallel, each isolated in its own git worktree. Each task gets its own branch, working directory and runs Claude Code session using Agent SDK That lets one agent write tests while another builds a feature or investigates a bug — all in the same repo without conflicts. We ended up building most of ChatML itself using this workflow and merged 750+ pull requests generated through parallel agent sessions. That experiment convinced me the model actually works. More about that experience building here: https://chatml.com/blog/we-built-entire-product-with-ai-750-... GitHub https://github.com/chatml/chatml Website https://chatml.com The project is open source and currently macOS-focused. I’d especially appreciate help from anyone interested in bringing it to Windows and Linux. If you’ve worked on cross-platform desktop apps, packaging, or filesystem/watchers issues across platforms, contributions or advice would be very welcome. We have it running but with a few bugs. For context, I previously co-founded KnowBe4, now co-founder of ReadingMinds.ai and for the past few years have explored developer tools around AI-assisted software engineering. Our engineers in several companies I am involved with are using ChatML and we have received great feedback, bugs and security issues. Happy to answer questions about the architecture, the git worktree model, or how parallel agent workflows feel in practice. I am using this tool for ALL of my development, and I am happy to share with the community. -Marcio
Does the same job
all alternatives →
- SRShikigami, run AI coding agents in parallel, each in a Git worktreeJul 2026 · shikigami.dev · ▲7
Hello all, I'm a software developer. Over the last few months more and more of my work has turned into using coding agents instead of typing the whole code myself. Usually a few claude sessions at once, sometimes codex, one per feature or per revealed bug. I ran them in a split terminal for a few weeks, and quickly spotted two main problems. The first is that I couldn't easily tell which agent was stuck waiting on me and which was still working, so I'd cycle through sessions and checking on them. The second one: agents sharing a single branch step on each other. Two of them could be editing…
More ai this month
the category →
I trained a 125M-parameter transformer to autocomplete piano performances in real time (~108 notes/sec on an iPhone 15). The idea is basically GitHub Copilot or Tabnine, except instead of prompting it with code, you prompt it by playing a few notes on a MIDI piano. The model then continues what you played, entirely on-device. The app is free if anyone wants to try it. Happy to answer questions about the model, training, Core ML, or the many things that didn't work.
AI · 17d ago · simedw.com
Astute▲585Automate your B2B brand going viral, with new media creators
AI · 18d ago · company-app.joinastute.com
Hey HN, Henry from Cactus here! We previously released Cactus Needle, a 14MB agentic LLM for tool call, device use, and structured extraction for phones, wearables, smart homes, small robots and microcontrollers. We got really great feedback here, and have now incorporated the suggestions to release Needle 2. The whole model is a single 14MB binary that runs a full session in 28MB of RAM; 45m parameters at 2bit compression. Needle hits 500 tokens/sec decode speed on a Raspberry Pi 5, sits between 400-1,500 tokens/sec on VR devices like Meta Quest 3S and Apple Vision Pro, and ranges…
AI · 27d ago · cactuscompute.com
Launched alongside, March 2026
the whole month →
Switch from ChatGPT to Claude with import memory feature
AI · Mar 2026 · claude.com
