Repository navigation
Conversation
…reen runContest takes an AbortSignal and records a cancelled contest; startPreview runs a candidate's ref in a scratch worktree; candidateDiff and history read back what the app shows. Models from unconnected integrations or disabled providers no longer take builder slots, and the consensus gate fails with fewer than two contenders.
The main process runs the contest library against the engine it already started and streams progress to the interface over the bridge (start, cancel, choose, apply, preview, diff, history).
The first tab of a project: task, live lanes, verdict against the repo's standard, gates, who-beat-whom matrix, preview, and names revealed only after choosing. en and ko.
Playwright against the unpacked Linux build under xvfb-run, with scripted vendors and the bundled engine. CI uploads the screenshots.
flxkm522
marked this pull request as draft
October 12, 2026 08:30
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Stack: #75 (shell) → #76 (contest engine) → #77 (Quality Lab) → this (M3, build-off screen). Base is
feature/quality-lab; retarget todevelopas the stack merges.What
The blind build-off is now the first tab of every project in Redrob Code Desktop.
Redrob-Model/Redrob-Provider/Redrob-Contesttrailers.cancelled.How
packages/contest:AbortSignalsupport,startPreview,candidateDiff,history. Models from unconnected integrations or disabled providers no longer take builder slots. The consensus gate fails when fewer than two contenders are left, so a one-model "build-off" can't pass.apps/desktop: the main process hosts the contest library against the engine it already runs (src/main/buildoff.ts) and streams progress over the bridge. There are no new engine routes; the/contestHTTP surface stays deferred until Cowork or a remote runtime needs one.apps/studio:views/BuildOff.tsx,lib/buildoff.ts, tests.Verification
apps/desktop/e2e/buildoff.mjsuses Playwright to drive a whole blind build-off through the packaged app's window. It runs with scripted vendors and the bundled engine, with no key and no network. Locally: 10/10. It checks that names stay hidden before choosing, that the judges' choice is the correct candidate, that applying creates a branch, and that the commit records the model.desktop (linux)runs it underxvfb-runand uploadsbuildoff-screenshots.bun turbo typecheck21/21; contest 27/27; desktop 24/24; studio 6/6;bun install --frozen-lockfileclean.Not verified