Quick Start Guide
Get started with Rampart in under 5 minutes
2. Filter User Input Before Sending to LLM
Check prompts for prompt injection, PII, and other threats
# Python example
import requests
response = requests.post(
"https://rampart.arunrao.com/api/v1/filter",
headers={"Authorization": "Bearer rmp_live_xxxxx"},
json={
"content": user_input,
"filters": ["prompt_injection", "pii"],
"user_id": "user_123"
}
)
result = response.json()
# Check if content is safe
if not result["is_safe"]:
print("⚠️ Threat detected:", result["threats"])
print("Details:", result["prompt_injection"])
else:
# Safe to send to your LLM
print("✓ Content is safe - proceed to LLM")
llm_response = call_your_llm(user_input)3. Or Use the Secured LLM Proxy (Optional)
Automatic security checks + LLM calls in one request
# Combined filtering + LLM call
response = requests.post(
"https://rampart.arunrao.com/api/v1/llm/complete",
headers={"Authorization": "Bearer rmp_live_xxxxx"},
json={
"prompt": user_input,
"model": "gpt-4",
"provider": "openai",
"user_id": "user_123"
}
)
result = response.json()
if result["blocked"]:
print("⚠️ Request blocked:", result["reason"])
else:
print("✓ LLM Response:", result["response"])
print("Security checks:", result["security_checks"])🎉 That's it! Your LLM calls are now protected
All requests are automatically scanned for prompt injection, data exfiltration, and policy violations.