> ## Documentation Index
> Fetch the complete documentation index at: https://docs.negentro.tech/llms.txt
> Use this file to discover all available pages before exploring further.

# Vector Databases

> Select and configure the high-dimensional vector substrate powering semantic memory recall.

## Vector Store Matrix

| Vector Store | Index Type | Best For | Connection Config |
| :- | :- | :- | :- |
| pgvector (Default) | HNSW / IVFFlat | Production apps using PostgreSQL; transactional safety. | pgvector://user:pass\@host:5432/db |
| Qdrant | HNSW with payload filters | Ultra-fast Rust vector engine with complex metadata filtering. | [http://localhost:6333](http://localhost:6333) |
| Chroma | HNSW (DuckDB/ClickHouse) | Local prototyping, embedded apps, zero external services. | path/to/chroma\_db |
| Milvus | IVF / HNSW / SCaNN | Enterprise scale across 100M+ vector records. | localhost:19530 |

## Configuring pgvector

```python pgvector_setup.py theme={null}
from negentro import Memory

config = {
    "vector_store": {
        "provider": "pgvector",
        "config": {
            "connection_string": "postgresql://postgres:secret@localhost:5432/negentro",
            "table_name": "agent_memories",
            "distance_metric": "cosine"
        }
    }
}

m = Memory.from_config(config)
```


This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.