Story

Hi HN!

I built Opaal because writing multi-agent orchestration prompts was becoming tedious and error-prone. Every time I wanted to coordinate 3-5 AI agents on a complex task, I would spend 20+ minutes crafting the prompt by hand.

Opaal (Orchestration Prompts for Agentic AI Launch) lets you design these workflows visually instead. You drag agent cards onto a canvas, organize them into phases (columns), draw connections between them, and the app generates a production-ready prompt automatically. The prompt updates live as you build.

Built with Electron + React + React Flow + Zustand + Tailwind CSS v4.

Key features: - 15 agent roles (Researcher, Architect, Developer, Reviewer, etc.) - Smart auto-connections between adjacent phases - Manual wiring for custom data flow - 3 starter templates (Code Review, Feature Build, Bug Fix) - Auto-detects installed Claude Code skills - Save/load .opaal files, export to CLAUDE.md - Full keyboard shortcuts, undo/redo, multi-select

MIT licensed. Would love feedback on what features would make this more useful for your workflows.