@agentprodready/vector-store-pgvector

pgvector adapter for @agentprodready/vector-store — durable semantic index for AgentProdReady Memory.

| | | |---|---| | Status | Production provider published (1.0.x) | | Install | npm install @agentprodready/vector-store-pgvector | | Module | ESM · Node.js >=22 <25 | | License | MIT |

When to use

You need Postgres + pgvector behind Memory semantic/hybrid search. Skip this for Simple ephemeral memory: true.

Install

npm install @agentprodready/vector-store @agentprodready/vector-store-pgvector

Requires DATABASE_URL (same shape as @agentprodready/persistence-postgres).

Sample

import { PgvectorVectorStore } from '@agentprodready/vector-store-pgvector';

const store = new PgvectorVectorStore({
  config,
  dimensions: 32,
  embeddingModelId: 'reference-embedding-32',
});
await store.assertReady();

VECTOR_INDEX_PROFILE=reference-32 pnpm db:migrate:vector
VECTOR_INDEX_PROFILE=reference-32 pnpm db:status:vector

| VECTOR_INDEX_PROFILE | Dimensions | Embedding model id | |---|---|---| | reference-32 | 32 | reference-embedding-32 | | openai-1536-small | 1536 | text-embedding-3-small |

assertReady fails closed unless the migrated schema contract matches dimensions / model / metric.

Ownership

| Owns | Does not own | |---|---| | Derived memory_vector_index schema, HNSW cosine index, PgvectorVectorStore | Memory authorization; Persistence generic repos; embeddings |

Docs

• Vector search guide
• Repository README

License

MIT © 2026 ameenmari