[Candidate / unverified] Solution for 5a99c35a-2493-41b6-9a94-55b49df813ac: Treat rebuild/fan-out like an explicit target set with a hard cardinality bound:

## Support semantics This is a **candidate / unverified** Solution from the KFA Solution Factory Tier-2 gate. It is source-grounded research material, not an Outcome, and does **not** claim independent reproduction or confirmed field success. ## Cause Unbounded fan-out occurs when rebuild/impact work is not constrained to an explicit target set. Primary sources show systems commonly require explicit selectors (Kubernetes labels) and that unscoped refresh can replace entire views (PostgreSQL materialized views)—supporting a bounded explicit-target control pattern. ## Steps 1. Model dependencies so a change resolves to a finite set of downstream targets (labels, keys, or shard ids). 2. Persist the target set and a max fan-out bound in the change receipt. 3. Enqueue rebuild work only for listed targets; route overflow to a slower convergence path. 4. Emit metrics/receipts naming target set size, bound, and overflow state. ## Expected result A single edit triggers work proportional to the declared impact set, not an implicit full-table or full-cache refresh. ## Misleading approaches - Global REFRESH or rebuild-all flags after every canonical edit. - Relying on implicit cache invalidation without enumerating keys. ## Limitations - A bound is a safety rail; it does not prove completeness of impact analysis. - Incorrect dependency graphs can omit required targets. - Not reproduced on a private deployment in this run. - Sources document selector/full-refresh primitives; they do not prescribe this exact factory bound-fan-out procedure. - Analogy across Kubernetes/PostgreSQL/DynamoDB pagination is illustrative, not a cross-product CVE fix. ## Unknowns - Optimal bound sizing without production traffic measurements. - Whether the original Problem environment already had an impact-target API that should be reused instead of a new bound. ## Evidence claims - (documented_platform_behavior) Kubernetes label selectors identify a set of objects and are used to filter/list/watch and to target pods (e.g. Services, controllers). — https://kubernetes.io/docs/concepts/overview/working-with-objects/labels/ - (documented_platform_behavior) PostgreSQL REFRESH MATERIALIZED VIEW completely replaces the contents of a materialized view. — https://www.postgresql.org/docs/current/sql-refreshmaterializedview.html