@postnesia/db

SQLite database layer for the Postnesia memory system. Provides schema migrations, Gemini embeddings, vector search, access tracking, and importance dynamics.

Exports

import { getDb, queries, createMemory } from '@postnesia/db';
import { embed } from '@postnesia/db/embeddings';
import { logAccess } from '@postnesia/db/access';
import { runConsolidation } from '@postnesia/db/importance';

Environment Variables

| Variable | Required | Default | Description | |---|---|---|---| | DATABASE_URL | Yes | — | Absolute file:// URL to the SQLite database | | GEMINI_API_KEY | Yes | — | Google Gemini API key for embeddings | | EMBEDDING_MODEL | No | gemini-embedding-001 | Embedding model name | | EMBEDDING_DIMENSIONS | No | 768 | Embedding vector dimensions |

DATABASE_URL must be an absolute path, e.g. file:///home/user/workspace/memory.db.

Setup

First-time (copies schema into your project and migrates)

npx postnesia-migrate-init

Apply pending migrations

npx postnesia-migrate-dev

Seed core memories

Seeds the operational guide into the database as a core memory:

npx postnesia-seed

Bins

| Command | Description | |---|---| | postnesia-migrate-init | Copy schema.prisma to your project and run initial migration | | postnesia-migrate-dev | Run prisma migrate dev against your database | | postnesia-seed | Insert core bootstrap memories |

Schema Overview

• memory — main memory records with content, content_l1, type, importance, core, timestamps
• tag / memory_tag — many-to-many tag associations
• relationship — directional links between memories
• vec_memories — sqlite-vec virtual table for vector similarity search
• access_log — read access history used by the importance system
• journal — daily journal entries
• task — task records with status and optional session grouping