AI Testing Environments

A multi-language workspace for testing AI providers (OpenAI, Anthropic, Google Gemini) with consistent patterns across Python, Node.js, and C++.

Overview

This repository provides reference implementations for integrating with major AI providers in three programming languages. Each environment features:

• Environment-based Configuration (development/production)
• Multi-Provider Support (OpenAI, Anthropic, Google Gemini)
• Unified Client Pattern (auto-selection with fallback)
• Cost Tracking & Limits

Structure

``` ai_testing/ ├── README.md # This file ├── python/ # Python environment │ ├── venv/ # Virtual environment │ ├── requirements.txt # Dependencies │ ├── config/ # Configuration system │ ├── clients/ # AI provider clients │ ├── frameworks/ # LangChain, LlamaIndex integrations │ ├── storage/ # Vector stores (ChromaDB) │ ├── tracking/ # Usage tracking │ ├── tests/ # Test suite │ └── docs/ # Documentation ├── nodejs/ # Node.js environment │ ├── package.json # Dependencies │ ├── src/config/ # Configuration system │ ├── src/clients/ # AI provider clients │ └── src/index.js # Entry point └── cpp/ # C++ environment ├── CMakeLists.txt # Build configuration ├── include/ # Header files │ ├── config/ # Config structures │ └── clients/ # Client interfaces └── src/ # Source files ```

Quick Start

Python

```bash cd python source venv/bin/activate python -c "from clients.ai_clients import UnifiedClient; print(UnifiedClient().chat('Hello!'))" ```

Node.js

```bash cd nodejs npm install node src/index.js ```

C++

```bash cd cpp mkdir build && cd build cmake .. && cmake --build . ./ai_testing ```

Environment Variables

All three environments use the same environment variables:

```bash

API Keys

export OPENAI_API_KEY="sk-..." export ANTHROPIC_API_KEY="sk-ant-..." export GEMINI_API_KEY="..."

Environment selection

export ENVIRONMENT=development # or 'production' ```

Configuration Comparison

Setting	Development	Production
OpenAI Model	gpt-3.5-turbo	gpt-4o
Anthropic Model	claude-3-haiku	claude-sonnet-4
Google Model	gemini-1.5-flash	gemini-1.5-pro
Daily Cost Limit	$5.00	$100.00
Cache TTL	24 hours	1 hour
Log Level	DEBUG	INFO
Max Tokens	512-1024	4096-8192

Common Patterns

Unified Client

All environments implement a `UnifiedClient` that automatically selects an available provider:

```python

Python

from clients.ai_clients import UnifiedClient client = UnifiedClient() response = client.chat("Hello!") ```

```javascript // Node.js import { UnifiedClient } from './clients/ai-clients.js'; const client = new UnifiedClient(); const response = await client.chat("Hello!"); ```

```cpp // C++ clients::UnifiedClient client; auto response = client.chat("Hello!"); ```

Config Access

```python

Python

from config import get_config config = get_config() print(config.models.openai_default) ```

```javascript // Node.js import { getConfig } from './config/index.js'; const config = getConfig(); console.log(config.models.openaiDefault); ```

```cpp // C++ auto cfg = config::get_config(); std::cout << cfg.models.openai_default << std::endl; ```

Features by Language

Feature	Python	Node.js	C++
OpenAI SDK	✓	✓	REST
Anthropic SDK	✓	✓	REST
Google SDK	✓	✓	REST
LangChain	✓	-	-
LlamaIndex	✓	-	-
Vector Store	✓ (ChromaDB)	-	-
Usage Tracking	✓	Basic	Basic
Key Rotation	✓	-	-

Documentation

• Python Environment
• Node.js Environment
• C++ Environment

Use Cases

This workspace is designed as a reference implementation for:

1. Learning - Understanding AI API integration patterns
2. Prototyping - Quick testing of AI features
3. Comparison - Evaluating providers and languages
4. Templates - Starting point for new projects

License

Open Source - Free to use, modify, and distribute.