What problem does it solve? Systems keep multiple copies of the same fact across database, cache, server memory, client state, and URL, and bugs arise when these layers disagree. This Skill guides an agent to systematically find where layers diverge and where the system acts on stale or wrong state. ## Core Features & Use Cases - Layer Inventory and Truth Mapping: Enumerate which state layers exist in a flow, label the source of truth for each fact, and trace write and read paths across copies. - Concrete Desync Testing: Test stale cache reads, read-replica lag, un-reverted optimistic UI, URL state that the server no longer recognizes, and multi-client divergence. - Evidence-Graded Findings: Report each divergence with confidence levels (CONFIRMED, HIGH CONFIDENCE, POSSIBLE, SPECULATIVE) and remediation strategies such as write-through caching, synchronous invalidation, read-your-writes, and optimistic rollback. - Use Case: After users report that a balance shows 100 in the UI but 50 in the database, use this Skill to trace the missing cache invalidation and produce a reproducible finding. ## Quick Start Audit this checkout flow for state consistency issues across the database, Redis cache, and client state, and report any divergences with reproduction steps.
