Skills, templates, and guides for building with AI agents — from a product person, not an engineer.
I'm Dave Killeen. Field CPO at Pendo.io by day, building Dex on the side. I can't write TypeScript from scratch, but I can plan, I can talk to AI, and I can orchestrate agents to ship real software while I sleep.
This repo is where I share the tools and methods that make that possible.
Reusable instruction files that teach AI coding agents how to do specific jobs well.
- Agent-Oriented PRD Writer — An interactive wizard that helps you write specs optimized for AI agent execution. Includes a 4-agent review swarm that stress-tests your spec before any code gets written.
I use:
- Claude Code (10X Max subscription)
- Agent Flywheel (free, open source) by Jeffrey Emanuel
- A cloud VPS (OVH, ~£63/month)
The flywheel spawns multiple AI agents in parallel on the VPS. Each agent gets a work package from the spec. They run autonomously. I check in from my phone.
Most AI coding content is written by engineers, for engineers. I'm not one. I'm a product person who figured out that the planning skills I already had — writing clear specs, defining success criteria, thinking in user outcomes — are exactly what makes AI agents perform well.
If you're in product, design, or leadership and wondering whether this is for you: it is.
MIT — use whatever helps you.