A focused suite of CI tools for studios that need actionable build and repository signals—not another dashboard to babysit.
Candidate compared with main@8f32c1
Small enough to adopt quickly. Specific enough to tell you exactly what changed.
Turn opaque Unity build changes into a reviewable report: size, assets, shaders, bundles, and build duration.
View pilot brief ↗Catch broken GUIDs, missing meta files, LFS mistakes, and dependency drift before the branch lands.
View pilot brief ↗Compare Android and iOS candidates for permissions, SDKs, native libraries, symbols, and package growth.
View pilot brief ↗Five deterministic browser-local evaluators test whether the narrow output is useful, already covered, impractical, noisy, or not actionable before a private pilot begins.
Compare accepted and candidate Unity build summaries under an exact build-contract gate with visible coverage, budgets, and dependency movement.
Would the ranked movement change your build review?Build summaries stay in the browser · missing evidence cannot silently passRun the local evaluator ↗Compare accepted and candidate Unity repository manifests under one exact extraction contract with visible meta/GUID, serialized-reference, package, and Git LFS evidence.
Would the structural evidence change your pull-request review?Repository manifests stay in the browser · incomplete evidence cannot silently passRun the local evaluator ↗Compare accepted and candidate Android or iOS release inventories under an exact package contract with visible expected changes, evidence coverage, and payload budgets.
Would the package movement change your release review?Inventories stay in the browser · no package upload or store-clearance claimRun the local evaluator ↗Compare a failed Unity CI log with a last-green run and receive one bounded next check only when the context is comparable.
Did the comparison change what you would check next?Logs stay in the browser · manual redaction review requiredRun the local evaluator ↗Compare repeated editor-loop timing summaries under exact environment, sample-count, spread, and visible budget gates.
Would the timing verdict change your review?Timing summaries stay in the browser · mismatch becomes unknownRun the local evaluator ↗Practical review contracts, grounded in primary Unity and platform documentation, with an explicit line around what the evidence cannot prove.
We are opening a small number of design-partner pilots. The clock starts after the evaluator, safe input, exact output, transfer method, and success criterion are confirmed.
Discuss your workflow ↗