HACKCOFFEEBEANS

Lumen — agent-powered end-to-end testing for everyone. E2E today means brittle selectors, hand-written tests, and stack traces when something breaks. We flipped the model. Point autoend at any web app. AI agents explore it like real users, discover meaningful flows (login, checkout, signup), and verify each one before it enters your Flow Map — a living baseline of what demonstrably works, committed as plain Playwright in your repo. It creates interactive graph with user journeys and video proofs. Every next Run replays the entire map first — headless, parallel, LLM-free, in seconds. Agents then spend budget only on new surface. When something breaks, you don't get a log wall: you get a beautiful interactive Report — verdict up top, tiered findings, timelines, and WebM evidence you can watch like a security camera. For everyone: no test framework to learn, no scripts to maintain. For teams: shared Flow Map in git, portable artifacts, dismiss/suppress from the UI. Tests that discover themselves. Regressions you can see. That's autoend.