> ## 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.

# Configuration

> Fine-tune every layer of the memory stack — from vector dimensions and chunking strategies to LLM inference pipelines.

## Complete Configuration Schema

Negentro reads configuration from negentro.config.json , a JavaScript/TypeScript config file, or environment variables:

```json negentro.config.json theme={null}
{
  "llm": {
    "provider": "openai",
    "model": "gpt-4o",
    "temperature": 0.1
  },
  "embedder": {
    "provider": "openai",
    "model": "text-embedding-3-small",
    "dimension": 1536
  },
  "vector_store": {
    "provider": "pgvector",
    "connection_uri": "postgresql://negentro:secret@localhost:5432/negentro",
    "collection_name": "negentro_memories"
  },
  "storage": {
    "bitemporal_enabled": true,
    "history_store": "postgres"
  },
  "cache": {
    "provider": "redis",
    "url": "redis://localhost:6379",
    "ttl_seconds": 3600
  }
}
```

## Resolution Precedence

Configuration values are resolved in the following priority order:

* Direct arguments passed to SDK methods (e.g. user\_id , limit ).
* System environment variables (e.g. OPENAI\_API\_KEY , DATABASE\_URL ).
* Configuration file ( negentro.config.json ).
* Default production settings.


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