Skip to content

amr080/super

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

16 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

[WARNING][WARNING]
[EDUCATIONAL PURPOSES ONLY]
[WARNING][WARNING]



base_url https://7541e4a8fa7a53898dab99d4125db4adef095535.replit.app/

API Endpoints
POST /infect: Receives infection data (host) and triggers further propagation.
GET /payload: Returns polymorphic worm code for infection.
GET /: Basic health check, returns "OK".
GET /*: Serves decoy HTML page with browser-based propagation.















[2025-02-22T03:37:01.276Z] Worm active on 169.254.8.1:3000



process.env.PORT || 3000







===============================================


• Containerize the app (Dockerfile that installs dependencies and exposes port)
• Deploy the container on a cloud platform (AWS, GCP, DigitalOcean, etc.)
• Use orchestration (Kubernetes/ECS) to manage multiple instances
• Set up a load balancer to distribute incoming HTTPS traffic
• Open necessary firewall/security group ports for public access
• Map a domain name to the load balancer’s public IP via DNS
• Configure auto-scaling rules based on traffic and resource usage
• Implement monitoring and logging for health and performance tracking









Creates HTTPS server on port 3000
Attempts to spread across network using HTTP, DNS, and P2P methods
Maintains list of infected hosts
Uses stealth techniques like:
Fake headers
Random user agents
Payload obfuscation
Anti-debugging
Includes browser-based propagation through WebRTC
Self-destructs after 1 hour













Multi-Vector Propagation
1 HTTP Flood: Parallel request saturation
2 DNS Tunneling: Hidden payloads in DNS queries
3 P2P Mesh: Direct host-to-host communication

Advanced Evasion
1 Dynamic Code Obfuscation: Real-time code mutation
2 Junk Code Injection: Random padding generation
3 Header Spoofing: Randomized legitimate-looking headers

Network Radar
1 Subnet Scanning: 3 simultaneous subnet probes
2 Po2rt Randomization: Dynamic target port selection
3 WebRTC Exploit: Internal network discovery

Persistence Mechanisms
1 Cross-Platform Autostart: Windows Task Scheduler & cron
2 Process Masquerading: Disguised as "SystemUpdate"
3 Anti-Debugging: Detects inspection tools

Performance Optimization
1 Connection Pooling: Reusable keep-alive sockets
2 Zero Sleep Architecture: No delays between attempts
3 Binary Storage: Faster infection list processing

Self-Preservation
1 1-Hour Decay: Automatic self-destruction
2 Encrypted Payloads: XOR-based code scrambling
3 Process Monitoring: Restarts if terminated


Testing Protocol:
1 Run in isolated VM network
2 Monitor with Wireshark/network tap
3 Use dedicated test machines
4 Never expose to public internet



SURF THE WEB
• Software is exposed by listening on a public IP/port and allowing inbound connections
• For a Node.js file (e.g., index.js), create an HTTP server listening on a port (e.g., 80 or 3000)
• Ensure any firewall/NAT port forwarding is configured to accept external traffic on that port
• Example:
npm init -y
Install express: npm i express
index.js:
js
Copy
Edit
const express = require('express');
const app = express();
app.get('/', (req, res) => {
  res.send('Hello, public internet!');
});
app.listen(3000, () => console.log('Listening on 3000'));
Run node index.js, then ensure port 3000 is accessible via your public IP
• Users can connect by visiting http://yourPublicIP:3000/ in their browser