requests-ai-validator 1.0.8

AI-powered validation wrapper for Python requests library

🤖 requests-ai-validator

AI-powered REST API testing framework - drop-in replacement for requests with intelligent validation

🎯 What is it?

requests-ai-validator is a drop-in replacement for the popular requests library that adds AI-powered validation capabilities to your REST API tests. It uses Large Language Models (LLMs) to intelligently validate API responses, request-response consistency, and business logic.

✨ Key Features

• 🔄 Drop-in replacement for requests - change only the import statement
• 🤖 AI-powered validation - intelligent analysis of API interactions
• 📊 Multiple AI providers - OpenAI, Anthropic, Ollama support
• 🎯 Schema validation - Pydantic models, JSON Schema, OpenAPI support
• 🌍 English feedback - clear and professional AI responses
• 📋 Business rules - custom validation rules
• 📊 Allure integration - detailed reporting
• ⚙️ Environment configuration - easy setup via .env

🚀 Quick Start

Installation

pip install requests-ai-validator

Basic Usage

# Replace this line:
import requests

# With this line:
import requests_ai_validator as requests

# Use exactly like regular requests + AI validation:
from pydantic import BaseModel

class UserModel(BaseModel):
    id: int
    name: str
    email: str

# Regular request
response = requests.get("https://api.example.com/users/1")

# With AI validation - just add parameters!
response = requests.get(
    "https://api.example.com/users/1",
    ai_validation=True,
    ai_schema=UserModel,
    ai_rules=["User must exist and be valid"]
)
# If AI finds issues → test fails with detailed feedback

⚙️ Configuration

Environment Variables (.env)

# AI Configuration
AI_TOKEN=your-api-key-here
AI_PROVIDER=openai
AI_MODEL=gpt-3.5-turbo

Supported Providers

Provider	Models	API Key
OpenAI	gpt-3.5-turbo, gpt-4, gpt-4-turbo, gpt-4o	AI_TOKEN
Anthropic	claude-3-haiku, claude-3-sonnet, claude-3-opus	AI_TOKEN
Ollama	llama2, codellama, mistral	No API key (local)

Language

All AI feedback is provided in English for universal compatibility and clarity.

📖 Usage Examples

1. Simple GET Request

import requests_ai_validator as requests
from pydantic import BaseModel

class UserModel(BaseModel):
    id: int
    name: str
    email: str

# AI validation built into the request
response = requests.get(
    "https://api.example.com/users/1",
    ai_validation=True,
    ai_schema=UserModel
)

2. POST Request with Business Rules

response = requests.post(
    "https://api.example.com/users",
    json={"name": "John", "email": "[e-mail]"},
    ai_validation=True,
    ai_schema=UserModel,
    ai_rules=[
        "User should be successfully created",
        "ID should be a positive number"
    ]
)

3. Session Usage

session = requests.Session(ai_provider="openai")

response = session.get(
    url,
    ai_validation=True,
    ai_schema=Model
)

4. Manual Configuration

# Override environment settings
requests.configure_global_ai(
    ai_provider="anthropic",
    api_key="your-claude-key",
    model="claude-3-sonnet"
)

🎯 AI Validation Categories

The AI validates 5 key areas:

Category	Description	Example
http_compliance	HTTP protocol validation	"Status code 200 correct for GET request"
request_validation	Request payload validation	"Request contains valid data: name, email"
response_structure	Response format validation	"Response has valid JSON structure"
schema_compliance	Schema adherence	"All required fields present in response"
data_consistency	Request-response consistency	"Common fields match: name, email"

❌ Error Handling

When AI validation fails, you get detailed feedback:

AssertionError: ❌ AI validation failed: Schema validation issues found.
{'schema_compliance': "Missing required field 'missing_field' in response", 'data_consistency': 'Request field name does not match response'}

🔧 Advanced Usage

Positive vs Negative Tests

# Positive test - expect AI to find no issues
response = requests.get(url, ai_validation=True, expected_result=True)

# Negative test - expect AI to find issues  
response = requests.get(url, ai_validation=True, expected_result=False)

Custom AI Instructions

response = requests.post(
    url,
    json=data,
    ai_validation=True,
    ai_schema=Model,
    ai_rules=["Business rule 1", "Business rule 2"],
    ai_instructions=["Focus on data consistency", "Ignore performance"]
)

Manual AI Validation

# Get response first
response = requests.get(url)

# Then validate manually
validation_result = response.validate_with_ai(
    schema=Model,
    rules=["Custom rules"],
    expected_success=True
)

# Print detailed feedback (optional)
response.print_validation_details()

📊 Integration with Existing Code

Minimal Changes Required

# Before:
import requests
response = requests.get(url)
assert response.status_code == 200
return Model(**response.json())

# After:
import requests_ai_validator as requests  # Only this line changed
response = requests.get(url, ai_validation=True, ai_schema=Model)  # Added AI validation
assert response.status_code == 200
return Model(**response.json())

Framework Integration

Works seamlessly with popular testing frameworks:

• ✅ pytest - automatic test failure on AI issues
• ✅ Allure - detailed AI feedback in reports
• ✅ unittest - standard assertions work
• ✅ Custom frameworks - minimal integration effort

🔍 How It Works

1. Make request - uses standard requests library
2. AI analysis - sends request/response to LLM for validation
3. Intelligent feedback - gets structured analysis of API interaction
4. Test control - fails test if AI finds issues (configurable)

🎯 Benefits

• 🧠 Intelligent validation - catches issues traditional testing misses
• 🚀 Easy adoption - minimal code changes required
• 🔍 Deep analysis - validates business logic, not just structure
• 📊 Rich feedback - detailed explanations of issues found
• ⚡ Fast integration - works with existing test suites
• 🌍 Multi-language - supports different feedback languages

📋 Requirements

• Python 3.8+
• requests library
• pydantic for schema validation
• AI provider API key (OpenAI, Anthropic) or local Ollama

🔗 Links

• GitHub: Source code
• PyPI: Package
• Issues: Report bugs
• Examples: See project/ directory for integration examples

📄 License

MIT License - see LICENSE file for details.

🎨 Created with

This framework was created with Вайб кодинг - AI-powered development approach that combines human creativity with AI assistance to build elegant and powerful solutions.

---

Transform your API testing with AI intelligence! 🚀