@dawn-ai/memory-pgvector

Supported Postgres and pgvector storage for shared long-term memory across Dawn application instances.

Use this when: You need to share vector memory across multiple Dawn application instances.

Install

pnpm add @dawn-ai/memory-pgvector pg

The adapter requires the pg client at runtime; the install command includes it explicitly.

Example

import { pgvectorMemoryStore } from "@dawn-ai/memory-pgvector"

const connectionString = process.env.DATABASE_URL
if (!connectionString) throw new Error("DATABASE_URL is required")

const store = pgvectorMemoryStore({
  connectionString,
  dimensions: 1536,
})

// During application shutdown:
await store.close()

Runtime and stability

@dawn-ai/memory-pgvector is a supported node-only application surface. It initializes the pgvector extension and tables lazily, so the database role needs the corresponding DDL and extension privileges. Calling close() ends a store-created pool. For an injected caller-owned pool, close() is a no-op; the caller must end the pool separately. Stored memory is plaintext application data. Updating a record preserves its existing embedding; re-embed changed semantic content before relying on vector retrieval.

Related

• pgvector Memory API reference — exact options and lifecycle.
• Long-term Memory — application configuration.
• Recall and Retrieval — hybrid search behavior.
• @dawn-ai/memory — shared memory contracts and ranking primitives.

Maturity and support

Dawn is pre-1.0, and its public surface can change. All publishable Dawn packages release together as a fixed group; review the @dawn-ai/memory-pgvector changelog and upgrading guide before upgrading. For support, use GitHub Discussions; report defects in GitHub Issues.

License

MIT. See the repository license.