google-file-search

# Google File Search

Comprehensive skill for implementing Google File Search API with Gemini models for Retrieval-Augmented Generation (RAG).

## Overview

Google File Search provides managed RAG capabilities through:
- Automatic document chunking and embedding generation
- Semantic search across multiple document types
- Metadata-based filtering for targeted retrieval
- Grounding citations showing source documents
- Persistent storage with file search stores
- Integration with Gemini 2.5 models

**Two Official SDKs are available:**
- **TypeScript/JavaScript:** `@google/genai` npm package
- **Python:** `google-genai` pip package

## CRITICAL: Use Official SDKs Only

Do NOT use manual REST API calls or deprecated packages. Always use the official SDKs.

### TypeScript/JavaScript
```bash
npm install @google/genai
```

### Python
```bash
pip install google-genai
```

## Security: API Key Handling

**CRITICAL:** Never hardcode API keys.

```typescript
// ✅ CORRECT - TypeScript
const ai = new GoogleGenAI({ apiKey: process.env.GOOGLE_API_KEY! });

// ❌ WRONG
const ai = new GoogleGenAI({ apiKey: 'sk-abc123...' });
```

```python
# ✅ CORRECT - Python
client = genai.Client(api_key=os.getenv("GOOGLE_API_KEY"))

# ❌ WRONG
client = genai.Client(api_key="sk-abc123...")
```

Get API keys from: https://aistudio.google.com/apikey

## TypeScript Implementation

### Initialize Client

```typescript
import { GoogleGenAI } from '@google/genai';

const ai = new GoogleGenAI({ apiKey: process.env.GOOGLE_API_KEY! });
```

### Create File Search Store

```typescript
const store = await ai.fileSearchStores.create({
  config: { displayName: 'my-knowledge-base' }
});
console.log(`Store created: ${store.name}`);
```

### Upload and Index Documents

**Use `uploadToFileSearchStore()` which uploads AND indexes in one operation.**

```typescript
async function uploadFile(storeName: string, filePath: string, filename: string) {
  let operation = await ai.fileSearchStores.uploadToFileSearchStore({