opensearch-code-authoringlisted

# OpenSearch code authoring Inspect the target OpenSearch version, existing mappings/settings, client builder, document shape, workload, and evaluation signals before drafting. Preserve existing naming and index lifecycle conventions; do not invent field names, analyzer behavior, or model dimensions. ## Construction contract - Make mappings explicit for fields that are searched, filtered, sorted, aggregated, or embedded. Match `text`/`keyword`, analyzer, doc-values, and vector dimension/space to demonstrated use. - Build Query DSL from validated field/operator inputs. Put non-scoring constraints in filter context, constrain expensive query forms, and use a stable pagination strategy for the requested depth. - For vector or hybrid search, keep embedding model, dimension, space type, `k`, candidate controls, and filtering strategy compatible. Tune recall and p99 together; do not claim either improves without measurement. - Treat mapping type, analyzer, or vector changes as migration work. Supply a versioned index, reindex, alias/blue-green cutover, and rollback approach when the existing lifecycle requires it. - Author artifacts and diagnostic/benchmark commands only. Do not mutate an index or cluster without separate authorization. Read [mapping construction](reference/kb/mapping-construction.md) for schema changes and [query construction](reference/kb/query-construction.md) for Query DSL. The installed `opensearch-query-review`, `opensearch-schema-review`, and `opensearch