Advanced AI Middleware

Project Overview

AI-Powered Threat Detection & Incident Response

This project enhances real-time cybersecurity using AI-driven threat detection. It passively monitors endpoint activity, detects potential threats, and provides immediate feedback to mitigate risks based on OWASP best practices.

Features

• Real-Time Monitoring: Continuously analyzes endpoint and API activity for anomalies and threats.
• AI-Driven Detection: Uses machine learning to detect both known and unknown attack vectors.
• Scoped AI Middleware: Applies AI logic only to designated routes (/private/api).
• Efficient Worker Processing: Offloads AI tasks to a Node.js Worker Thread for better performance.
• User-Friendly Dashboard: Enables security teams to monitor logs, review threats, and manage responses.
• Open-Source & Scalable: Built with Vercel AI SDK and Ollama for cost-efficient, modular, and developer-friendly AI inference.

Technologies Used

• Hono: For advanced routing and middleware capabilities.
• Ollama: Local AI workflows.
• Vercel AI SDK: Wrapper around popular AI models for easy integration.
• Worker Threads: For non-blocking AI task execution.
• TypeScript: Ensuring type safety and maintainability.

Installation

To install the middleware package, use any of the following commands:

# npm
npm i @trythis/hono-sansai

# pnpm
pnpm add @trythis/hono-sansai

# yarn
yarn add @trythis/hono-sansai

🚀 Setting Up AI Middleware with Hono

1️⃣ Hono API Server with AI Middleware

import { Hono } from "hono";
import { sansAiMiddleware } from "@trythis/hono-sansai";
import path from "path";
import { Worker } from "worker_threads";

const app = new Hono();
const worker = new Worker(path.resolve("src", "./worker.js"));

worker.on("error", (err) => console.error("Worker error:", err));

// Scoped AI Middleware - Applied ONLY to "/private/api"
app.use(
  "/private/api",
  sansAiMiddleware({
    worker,
    environment: "debug",
    onWorkerMessage: ({ msg }) => console.log("AI Response:", msg),
  })
);

// Sample Private API Route Utilizing AI
app.post("/private/api/analyze", async (c) => {
  const data = await c.req.json();
  return c.json({ message: "Processing request with AI...", input: data });
});

// Sample Public Route (No AI Middleware)
app.get("/public/info", (c) => c.text("This is a public API endpoint."));

export default app;

2️⃣ Worker Thread for AI Processing (worker.js)

Handles AI processing separately to prevent blocking the main event loop.

import { parentPort } from "worker_threads";
import { createGoogleGenerativeAI } from "@ai-sdk/google";
import { sansAIWorker } from "@trythis/hono-sansai";

const google = createGoogleGenerativeAI({ apiKey: process.env.GEMINI_API_KEY });
const model = google("gemini-1.5-pro");

async function handler(payload) {
  const response = await sansAIWorker({ model, payload });
  parentPort?.postMessage(response);
}

parentPort?.on("message", handler);

🌟 Why This Approach?

✅ Scoped AI Processing – Ensures AI runs only on relevant routes (e.g., "/private/api").
✅ High Performance – Uses Worker Threads to prevent blocking the main event loop.
✅ Best Open-Source AI Stack – Powered by Vercel AI SDK and Ollama for seamless AI integration.
✅ Developer-Friendly – Modular, scalable, and cost-efficient.

Contribution Guidelines

We welcome contributions from the community. Follow these steps:

1. Fork the Repository
2. Create a Feature Branch: git checkout -b feature-name
3. Commit Changes: git commit -m 'Description of feature'
4. Push to Fork: git push origin feature-name
5. Submit a Pull Request

License

This project is licensed under the MIT License. See the LICENSE.txt file for details.

Acknowledgments

Special thanks to the organizers of the SANS AI Cybersecurity Hackathon for providing a platform to innovate in cybersecurity.

For more details, visit our Devpost project page.