Integrate with Fragniti's AI
Harness the power of our hybrid AI system to detect phishing threats in real-time. Our REST API is designed for simplicity, speed, and accuracy, returning easy-to-parse JSON responses.
Getting Started
This documentation provides all the information you need to integrate with the Fragniti API. The API is built around standard HTTP protocols and returns JSON-formatted responses, making it easy to use with any programming language. Our core endpoint, /api/predict, allows you to submit a URL and receive a comprehensive analysis. Let's dive in.
Authentication
Simplicity is key. The current version of the Fragniti API is public and does not require an API key for access. You can make requests to our endpoints directly without any authentication headers. We may introduce API keys in the future to manage usage and provide tiered access.
Analyze URL Endpoint
This is the primary endpoint for analyzing a URL. You send a URL in a JSON payload, and the API returns a detailed analysis and verdict.
HTTP Request
POST https://auth.fragniti.com/api/predict
Request Body
The request body must be a JSON object with a single key, url, containing the full URL you wish to analyze.
Response Fields
status: A string, eitherphishingorlegitimate.prediction: A human-readable summary of the verdict.explanation: A concise, AI-generated reason for the verdict.final_verdict: A string, eithergoodorbad.correction_made: A boolean indicating if the local model's prediction was corrected by the Gemini API.
Retrain Model Endpoint
This endpoint triggers a background process to retrain the local XGBoost model. This is a non-blocking operation.
HTTP Request
POST https://auth.fragniti.com/api/retrain
Health Check Endpoint
A simple health check endpoint to verify that the API service is running and the model is loaded.
HTTP Request
GET https://auth.fragniti.com/api/health
# Example Request
curl -X POST https://auth.fragniti.com/api/predict \
-H "Content-Type: application/json" \
-d '{"url": "http://example-phishing.com"}'{
"status": "phishing",
"prediction": "This URL appears to be PHISHING.",
"explanation": "The domain contains suspicious keywords...",
"final_verdict": "bad",
"correction_made": true
}