Roadmap
Proof of concept first, positioning before product
Built in the open, with a handful of real products. The order below is the order we will ship.
The chain, end to end, on one real feature set
Feature list extracted from a real repo and its docs · scenario tree for one feature set, corrected by its owner · tests emitted as Playwright and as browser instructions, and run · the readable view with one drift and one new scenario awaiting approval.
Live feature list
Repo and task-tracker hooks keep the feature list current while features are being built, so scenario drafts exist before the merge request does.
Merge-request checks
Dependency analysis on every MR against the scenario tree: which scenarios a change touches, which tests to re-run, which breaks are real.
Coverage analytics and drill-down
Logic-path coverage over the tree, per feature and per flow, for the owners' view.
Rules we hold ourselves to
Four constraints
Glue open source, pay only for models. Runtime, browser driving and reporting already exist; the value is in the layers above them.
Framework-agnostic. The scenario tree is the product; Playwright, Selenium, Cucumber or an AI with a browser are outputs, and a team keeps the one it already has.
Decide value before building it all. If the map layer turns out to exist and work elsewhere, we stop.
Test on real products. Every milestone is proven on a live codebase with a real owner before it is called done.
Be one of the first products
We are picking a small number of teams for the first milestone. Tell us about yours.