Vector Database Schema

JSON Schema for vector store configuration

Vector Database Definition

Schema Definition
VectorDBConfig configures a vector database target for knowledge storage.
config
object

VectorDBConnConfig defines connection and table options for a vector database.

id
string
type
string

Example

vector_dbs:
  - id: filesystem_faststart
    type: filesystem
  - id: pgvector_local
    type: pgvector
    config:
      dsn: "{{ .env.PGVECTOR_DSN }}"  # Optional for pgvector
      ensure_index: true
      table: knowledge_chunks
  - id: pgvector_default
    type: pgvector
    config:
      # DSN omitted - uses global postgres database config
      ensure_index: true
      table: knowledge_chunks
      dimension: 1536
  • filesystem is ideal for quickstarts and tests and persists embeddings to .compozy/cache/<vector_db_id>.store when no config.path is provided, so quickstarts do not require extra configuration.
  • pgvector DSN is optional - when not provided, pgvector uses the global postgres database configuration, reducing configuration duplication.
  • Qdrant requires an explicit DSN.
  • config.ensure_index controls whether Compozy enforces pgvector index creation on startup.
  • Use config.auth to pass API keys or bearer tokens for managed vector databases.

Resources