Unleash 1,000 Codex Instances On Your Coding Project

I'm building Git Swarm: an open-source system for orchestrating large numbers of AI coding agents around GitHub issues and pull requests. github - https://github.com/vukrosic/gitswarm join community to learn how to use AI to code and automate - https://www.skool.com/agentic-coding-lab-1713/about The goal is simple: let many agents claim issues, review PRs, launch terminals, use cheaper/faster models like MiniMax or GPT-5.4 Mini for implementation, and let humans supervise the architecture and product direction. In this demo, I show the early UI, issue claiming, review/merge agents, live terminals, and the next step: making Codex control the whole dashboard through CLI commands so software development can become much more automated. 0:00 Building a system for massive AI agent development 0:26 Review and merge agents 0:58 Git Swarm: agents building projects on GitHub 1:11 Tool overview: GitHub issues and PRs 1:38 Claiming issues with Codex Cloud or MiniMax 1:50 Why MiniMax is useful for cheap fast implementation 2:13 Open-source goal for companies and developers 2:31 Starting many agents in parallel 2:48 Automating supervision, reviews, and PR creation 3:13 UI needs and preventing duplicate claims 3:29 Small models for fast iteration 3:57 CLI control for Codex and Claude Code 4:30 Supporting multiple repositories and projects 4:55 Sidebar tabs for issues, PRs, and terminals 5:13 Single-file UI iteration vs refactor question 5:40 Mini implements the UI change 6:01 Claimed issue behavior and terminal controls 6:31 Finding and fixing a claim-state bug 6:36 Launching multiple claimed agents 7:00 Clone it, learn from it, or adapt the workflow