A public scout-powered regression evidence hub for Medusa's admin dashboard. 15 admin scenarios captured as scout-runnable Python; replayed against pinned Medusa releases; diff reports published here. Operated by BoxProbe — independent analysis, not affiliated with the upstream project.
Medusa is an open-source headless commerce platform. Its admin dashboard is a single-page app that drives a JSON API under the hood — exactly the kind of surface where API behavior can drift across releases while the UI stays visually identical. We maintain a scout-runnable suite covering the admin's core workflows (auth, customers, inventory, orders, products, campaigns, ...) and re-run it on each release pair we publish.
Two domain models (ProductCategory, Collection) gained an external_id field. Additive only — UI unchanged — but visible to integrations doing strict response-schema validation.
We re-run the suite as Medusa publishes new releases. Latest comparison gets the most attention; earlier ones stay archived here for cross-version analysis.
Test code lives in github.com/boxprobe/scout-medusa. Same scenarios are exercised across every version comparison.
Every report on this hub is reproducible. The scout-medusa demo repo includes Docker compose for both Medusa versions plus all 15 scenarios — clone, compose up, run scout twice, diff.
Operated by BoxProbe. Independent analysis — not affiliated with the upstream Medusa project unless noted. Reports here are not "verified" or "certified" by Medusa maintainers. If you maintain Medusa and would like a correction or removal, email [email protected] — we honor takedown requests within 7 days.