KAGAN - AI Orchestration Layer for Coding Agents

Hello ProductHunt! 👋 Launching Kagan today - a terminal UI designed for developers who want AI agents that augment their workflow, not replace their thinking. Why we built this: At MakerX, we were experimenting with Claude Code and OpenAI Code for rapid prototyping. The problem became obvious: agentic cli tools are powerful, but orchestrating them is painful. You're alt-tabbing, losing context, manually reviewing output, and second-guessing whether to merge. We realized the real work wasn't the code—it was the collaboration structure. The core insight: AI agents excel at repetitive, scoped tasks. Humans excel at judgment and complexity. What's missing is the interface that bridges them. Kanban + AI = that interface. What you get: 🎫 Terminal-based kanban — Organize agentic CLI sessions as discrete tickets 🤖 Autonomous mode — Low-risk tasks run fully AI-driven, review agent validates before merge 👥 Pair mode — Collaborate inside tmux with an AI agent in real-time 🔍 Smart planner — Convert complex requirements into scoped, actionable tickets using ACP + your preferred agent Who this is for: - Indie developers experimenting with AI-assisted workflows - Teams running exploratory projects (non-production codebases) - Anyone frustrated with context-switching between multiple agentic CLIs - Developers who want AI speed with human judgment Real use case: You have a 10-ticket sprint. Three are experimental. You put them in autonomous mode, Kagan's AI runs them, review agent flags issues, you merge when confident. No context switching. No manual orchestration. Open-source under MIT. Try it at kagan.sh. What's your biggest blocker with agentic tools right now? Is it speed, trust, or context management? Really curious what friction you're hitting. ⭐ GitHub star would mean a lot if you find this useful: https://github.com/aorumbayev/kagan