UX-OS / Orchestration
Proposed set: may be kept, renamed or dropped
Running the other sets against a real target until it meets a bar.
The other sets are methods. This one runs them against a real page and keeps going until it holds up.
What's missing: Only /crit works outside Workshopr.
Interviews you about a page, then critiques it against your own written rules and loops until a verifier passes it.
Testing: A gate with 55 test cases, and two headless runs.
Needs Node 18 or later and Playwright with Chromium.
git clone https://github.com/bbulman/critique-partner.git
cd critique-partner && ./install.shA self-looping designer: critique, fix, re-render and re-score until an exit bar is met, then learn from the run.
Testing: An independent critic scores every pass.
Runs only on Workshopr, so there is nothing to install.
Interviews you into a spec, a design brief and build slices, then runs the feature pipeline.
Runs only on Workshopr, so there is nothing to install.
Builds one slice of a planned feature and loops it until lint, tests, the design system and review all pass.
Testing: An independent reviewer scores every pass.
Runs only on Workshopr, so there is nothing to install.
The states are from my setup on 9 November 2026. Yours start from whatever you install.