llm-wiki/docs/superpowers/specs/2026-07-18-ca-skeleton-frontend-operational-contract-review/lanes/lane-02-benchmark-patterns.md

Lane Inventory

Path	Status	Evidence Lines	Extracted Facts
raw/project-notes/ca-skeleton-operational-contract.md	READ_FULL	L1-2226	frontend가 재사용할 schema와 backend 고유 세부를 분리

Evidence Matrix

Path	Status	Evidence	Extracted facts
raw/project-notes/ca-skeleton-operational-contract.md	READ_FULL	L26-48, L796-916, L962-985, L1177-1227, L1992-2115	goal/non-goal, registry, failure matrix, work-item contract, acceptance/readiness, evidence grade tracker

Reusable benchmark patterns

Pattern	Evidence	Frontend adaptation
invariant-oriented goal	L26-38	어떤 feature도 같은 API/error/state/dependency contract를 사용
registry row + owner + test	L796-897	route/env/storage/error/query-key/telemetry/release registry
scenario + expected failure	L899-916	happy/degraded/recovery/removal fixture matrix
Decision/Allowed/Forbidden/Test/Failure	L962-985	FE-OC contract schema
universal acceptance gate	L1177-1200	branch completion 전에 10개 binary 질문
evidence-grade tracker	L1992-2005, L2110-2115	planned → implemented → local → prod evidence path

Do not copy

• Gradle/Spring/JPA/Kafka/DB transaction, JVM/container lifecycle, backend-owned response status 정책.
• Frontend는 envelope를 소비·검증하고 browser/runtime/cache/chunk/config/render/storage 실패를 소유한다.

Counterargument

기준 문서의 분량과 registry row 수를 그대로 복제하면 frontend prospective contract에 불필요한 무게가 생긴다. 비교 기준은 줄 수가 아니라 stable owner, failure semantics, acceptance/evidence closure다.