Hey Product Hunt 👋 I built testcat because mobile UI testing is still either brittle scripts you babysit forever, or manual tapping before every release. testcat lets you describe a test in plain language and hands it to an AI agent that drives real iOS simulators — booting devices, tapping, typing, verifying — while you watch the whole thing live. At the end you get a verdict backed by actual actions and observations, not a hopeful "looks good." A few things I cared about: • Bring your own agent — Claude Code, Codex, opencode, or fully-local Ollama. • Trust by default — a completion gate cross-examines every success claim. No executed actions, no pass. • Local-first — everything persists to a local SQLite dashboard; nothing leaves your Mac. • Real devices too — it signs and installs an XCUITest runner so the same agent can drive your physical iPhone. It's open source (MIT) and macOS/Apple Silicon. Would love your feedback — especially on the agent profiles and the evidence/verdict model.

testcat

AI agents test your iOS app while you watch live

AI agents test your iOS app while you watch live