Quick Start
Get started in 5 minutes
1. Installation
# Install the Waclaude Python SDK pip install waclaude # Or using npm for Node.js npm install @obscurelabs/waclaude
2. Authentication
Get your API key from the ObscureLabs dashboard and set it as an environment variable:
export WACLAUDE_API_KEY="your-api-key-here"
3. Basic Usage
Replace your existing LLM calls with Waclaude-protected endpoints:
import waclaude
# Initialize the client
client = waclaude.Client()
# Secure LLM call with guardrails
response = client.complete(
messages=[{"role": "user", "content": "Help me write a secure function"}],
model="claude-3-sonnet",
guardrails=["prompt_injection", "secret_detection"]
)
print(response.content)Core Features
Comprehensive protection for your AI applications
Prompt Injection Detection
Advanced ML models detect sophisticated prompt injection attempts in real-time.
client.complete( messages=messages, guardrails=["prompt_injection"], injection_threshold=0.8 )
Secret Scanning
Continuous monitoring for exposed API keys, passwords, and sensitive data.
client.complete( messages=messages, guardrails=["secret_detection"], redact_secrets=True )
Content Filtering
Customizable content policies for harmful, biased, or inappropriate content.
client.complete( messages=messages, guardrails=["content_filter"], policy="enterprise" )
Configuration
Customize Waclaude for your use case
Environment Variables
# Required WACLAUDE_API_KEY=your-api-key # Optional WACLAUDE_ENDPOINT=https://api.obscurelabs.com/waclaude WACLAUDE_TIMEOUT=30 WACLAUDE_RETRY_ATTEMPTS=3
Guardrail Configuration
Configure detection thresholds and policies for each guardrail:
config = {
"prompt_injection": {
"enabled": True,
"threshold": 0.7,
"action": "block" # or "warn"
},
"secret_detection": {
"enabled": True,
"patterns": ["api_key", "password", "token"],
"redact": True
},
"content_filter": {
"enabled": True,
"policy": "strict",
"categories": ["harmful", "bias", "nsfw"]
}
}
client = waclaude.Client(config=config)Error Handling
Handle security violations and API errors gracefully:
try:
response = client.complete(messages=messages)
except waclaude.PromptInjectionError as e:
print(f"Prompt injection detected: {e.details}")
except waclaude.SecretDetectedError as e:
print(f"Secret found and redacted: {e.secret_type}")
except waclaude.ContentViolationError as e:
print(f"Content policy violation: {e.category}")Additional Resources
Explore more Waclaude documentation
API Reference
Complete API documentation
Python SDK
Python integration guide
Node.js SDK
JavaScript/TypeScript guide
Examples
Code examples and tutorials
Migration Guide
Migrate from existing solutions
Best Practices
Security recommendations
Need Help?
Our team is here to help you integrate Waclaude successfully. Reach out if you have questions or need assistance.