These browser-local evaluators expose the narrow deterministic core behind five pre-launch private-pilot offers. They test whether a bounded output changes a real workflow decision before any integration, account, package, repository, or technical-input transfer is discussed.
One accepted Unity build summary and one candidate summary under the same declared build contract.
OUTPUT
Comparable size, payload, shader, Addressables, duplication, duration, and dependency movement with visible coverage and budgets.
FAIL-CLOSED GATE
Ten build-contract fields must match and both core BuildSummary signals must be finite. Missing optional evidence stays review; malformed or mismatched evidence returns unknown.
Would the ranked movement change what the evaluator inspects or approves—or is this already solved, evidence-blocked, or not actionable?
One accepted and one candidate structured repository manifest produced under the same Unity, path, extractor, Package Manager, and Git LFS contract.
OUTPUT
Asset/meta parity, GUID identity, changed serialized-reference, package movement, and LFS storage findings with evidence, confidence, expectations, and coverage.
FAIL-CLOSED GATE
Ten repository-contract fields must match. Incomplete evidence stays bounded, malformed evidence returns unknown, and only definite observed contradictions block by default.
Would the structural evidence change what the repository owner inspects or approves—or is this already solved, evidence-blocked, or not actionable?
One accepted and one candidate Android or iOS release inventory produced under the same package and inspection contract.
OUTPUT
Traceable permission, entitlement, dependency, architecture, metadata, symbol, debug-content, and payload movement with expected-change labels.
FAIL-CLOSED GATE
Ten release-contract fields must match and core artifact/inventory surfaces must be valid. Optional evidence cannot silently pass; contract mismatch returns unknown.
Would the package movement change what the release owner inspects or approves—or is the workflow already solved, evidence-blocked, or not actionable?
One failed Unity CI log, one comparable last-green log, and optional changed paths.
OUTPUT
A stable fingerprint, first divergent error chain, comparability result, and one bounded next check.
FAIL-CLOSED GATE
Known sensitive patterns receive a narrow local preflight, but complete manual review remains mandatory. A Unity-version or target mismatch returns unknown.
Did the output change what the evaluator would check next—or is this already solved, input-blocked, or not actionable?
Reference and candidate timing-only summaries for one declared Unity editor-loop fixture.
OUTPUT
Phase medians, spread, absolute and relative deltas, visible budget status, and one bounded verification step.
FAIL-CLOSED GATE
Six environment fields must match, every phase needs five samples, and each sample set must stay within the visible spread ceiling. Otherwise the complete verdict is unknown.
Would the verdict change the evaluator's review—or are stable repeated measurements already covered, too noisy, or too expensive?
These are pre-adoption technical evaluations with no buyer-use evidence yet. A local result, page visit, or structured dropdown does not establish product demand.
After a non-incomplete result, the optional response form is isolated from the local input and output. XeSoft verifies the studio relationship and evaluates the non-confidential workflow text before treating a response as evidence.
A pilot begins only after the evaluator, safe input, exact output, execution boundary, and success criterion are confirmed. XeSoft can prepare the first testable integration within one week after that fit and scope are agreed.