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

# The Repository Pattern

> Understand the core data abstraction in FluxCRUD

## Overview

The `Repository` class is the heart of FluxCRUD's data layer. It wraps SQLAlchemy operations into a clean, type-safe API that handles:

* **Type Validation**: Using Pydantic schemas.
* **Caching**: Automatic read-through and write-through caching.
* **Batching**: Optimized database writes.
* **N+1 Prevention**: Optional DataLoader support (opt-in).

## Basic Usage

By default, the Repository uses direct SQL queries for maximum performance.

```python theme={null}
# Create
repo = Repository(session, model=User, schema=UserSchema)
user = await repo.create(UserCreate(name="Alice"))

# Read (Direct SQL)
user = await repo.get(1)
```

### Enabling DataLoader (N+1 Protection)

When resolving relationships in loops (e.g., GraphQL or nested API responses), enable the DataLoader to batch queries automatically.

```python theme={null}
repo = Repository(..., use_loader=True)
# Queries will be batched
user = await repo.get(1)
```

## Advanced Features

### Caching

When `CacheManager` is provided, `get()` calls check the cache first.

```python theme={null}
repo = Repository(..., cache_manager=redis_cache)
# This will hit Redis first, then DB if missing
user = await repo.get(1)
```

### Batching

Use the `batch_writer` context manager for high-performance inserts.

```python theme={null}
async with repo.batch_writer() as writer:
    for item in big_list:
        await writer.add(item)
```

### Learn More

* [Plugin System](/advanced/plugins) - Extend behavior with hooks.
* [Unit of Work](/advanced/unit-of-work) - Atomic transactions.
* [Streaming & Optimization](/advanced/streaming) - Handling large datasets.