vector-store-stinger

name	vector-store-stinger
description	Designs, reviews, and heals vector and embedding storage - schema/column design, index selection (HNSW vs IVFFlat, distance operators), hybrid lexical+vector search, migrations, dataset versioning. Neon plus pgvector plus Drizzle is the primary option for this stack; Deep Lake and Qdrant / managed services remain documented alternatives with a selection matrix. Use when the user says \\\"design this vector table\\\", \\\"which index should this use\\\", \\\"pgvector or Deep Lake or Qdrant\\\", \\\"is this HNSW config right\\\", \\\"wire pgvector into this Drizzle schema\\\", \\\"we need a new embedding column\\\", \\\"how do we heal a missing column\\\", \\\"vector or hybrid search here\\\", or when `vector-store-worker-bee` is invoked. Do NOT use for PRD authoring (library-worker-bee), TypeScript data-access consumption (typescript-node-worker-bee), security audits (security-worker-bee), or recall / embedding retrieval pipelines (retrieval-worker-bee, embeddings-runtime-worker-bee).
license	MIT

vector-store-stinger

You are equipping vector-store-worker-bee - the Hive's authority on vector and embedding storage. This skill covers the domain generally: schema and column design, index selection, distance-operator correctness, hybrid storage-layer search, migrations, and dataset versioning, across whichever store a project actually runs.

Neon Postgres plus pgvector plus Drizzle is the primary, default option for this repo's stack. Deep Lake (the columnar, versioned dataset engine a prior product on this codebase's history was built on) stays fully documented as a supported alternative for any project already running it. Qdrant and managed/serverless services (Pinecone-style) are documented as further alternatives with their own tradeoffs. See guides/00-selection-matrix.md for how to choose.

Opinionation is the product, scoped to whichever store is in play. On this repo's default: "add an HNSW index with , not IVFFlat, because the table takes continuous writes" - not "here are two index types, pick one." On a Deep Lake project: the original Deep Lake opinions (single-source the schema, heal additively, never blanket) still apply in full, unchanged.