Hey Product Hunt! 👋 I built Waterwheel after watching AI coding agents get faster and faster while QA stayed painfully manual. The agent can write a feature in minutes — why am I still clicking through a browser to verify it? Waterwheel solves the verification half of that equation. It reads your test files, drives a real browser, and reports back — all autonomously. Combined with Claude Code, I shipped a complete user auth feature with zero human intervention for $0.01 in test costs. What makes it different: - Tests are plain Markdown files — anyone on the team can write them - Optimized for token efficiency — high product quality without excessive API costs - Multi-provider support — you choose your AI and control your costs - Permission-layered by design — the agent can only access what you explicitly allow - Highly customizable — configurable for individual use cases and adaptable beyond QA - Pairs with Claude Code for fully autonomous TDD: code agent implements, Waterwheel verifies, no human in the loop The 1.0.0-mac release is free for local development. Try the demo project to see the full TDD loop in action: https://github.com/taodong/tdd-demo The full walkthrough (TDD with Claude Code + Waterwheel ) of our demo project: https://medium.com/@tao.dong_spr... Happy to answer any questions below! 🌊

Waterwheel Agent

AI browser test agent — plain English, zero QA bottleneck

AI browser test agent — plain English, zero QA bottleneck