https://github.com/HackerBlazeX/Offensive-AI-Attack-Path-Visualizer
🛡️ AI-powered offensive security framework that visualizes attack paths using multi-tool recon, risk scoring & OWASP mapping. Built for legal security testing & learning.
https://github.com/HackerBlazeX/Offensive-AI-Attack-Path-Visualizer
ai-security ai-security-tool bug-bounty-hunting cybersecurity llama-cpp-python owasp-top-10 penetration-testing-framework red-teaming streamlit-webapp
Last synced: about 1 month ago
JSON representation
🛡️ AI-powered offensive security framework that visualizes attack paths using multi-tool recon, risk scoring & OWASP mapping. Built for legal security testing & learning.
- Host: GitHub
- URL: https://github.com/HackerBlazeX/Offensive-AI-Attack-Path-Visualizer
- Owner: HackerBlazeX
- License: mit
- Created: 2026-01-30T12:29:09.000Z (4 months ago)
- Default Branch: main
- Last Pushed: 2026-02-01T22:12:54.000Z (4 months ago)
- Last Synced: 2026-02-02T05:48:09.639Z (4 months ago)
- Topics: ai-security, ai-security-tool, bug-bounty-hunting, cybersecurity, llama-cpp-python, owasp-top-10, penetration-testing-framework, red-teaming, streamlit-webapp
- Language: Python
- Homepage:
- Size: 10 MB
- Stars: 0
- Watchers: 0
- Forks: 0
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
- License: LICENSE
Awesome Lists containing this project
- awesome-ai-offensive-security - Offensive-AI-Attack-Path-Visualizer - A Windows-first offensive security framework that correlates recon signals, applies AI reasoning via local LLM, and generates realistic attack paths in an analyst-friendly dashboard. (Pentest & Red Teaming Agents)
README
🛡️ Offensive AI – Attack Path Visualizer
💻 Developed by Dip Kar (HackerBlazeX) 🚀
☕ Just give a domain — relax & sip your coffee.
🤖 Offensive AI handles yours 80% of the web penetration testing task
📄 and delivers a ready-to-use report with AI-driven insights.
AI-powered offensive security framework that transforms raw recon data into realistic, prioritised attack paths using local AI reasoning.
⚠️ For legal & authorised security testing and educational purposes only.
---
## 🚀 What is Offensive AI?
**Offensive AI – Attack Path Visualizer** is a Windows-first offensive security framework designed to help security professionals **think like a real attacker**, not just collect tool outputs.
Instead of showing scattered scan results, this framework:
- Correlates recon & scan signals
- Applies AI-driven reasoning using a **local LLM (llama.cpp)**
- Generates **realistic attack paths**
- Presents everything in a clean, analyst-friendly dashboard
Built for **pentesters, red teamers, bug bounty hunters, and cybersecurity learners**.
---
## 🧠 The Problem It Solves
Traditional penetration testing often suffers from:
- Too many tools, too much noise
- Disconnected findings
- Manual decision-making fatigue
- Difficulty deciding *what to exploit next*
**Offensive AI** bridges this gap by converting **raw technical data into structured offensive intelligence**.
---
## 🧠 How the Framework Works
The framework begins by collecting **raw signals** from multiple reconnaissance and scanning tools, such as:
- DNS resolution & reachability (nslookup, ping)
- Open ports & exposed services (Nmap)
- Subdomains (Subfinder)
- Live HTTP services, status codes & technologies (httpx)
- Known misconfigurations & CVE templates (Nuclei)
- Parameterised URLs (ParamSpider)
- Directory & file discovery (FFUF / Dirsearch)
- Web server misconfigurations (Nikto)
- Input-based testing signals (SQLMap, XSStrike)
All outputs are captured **as raw text**, without modifying or exploiting the target.
✔ No blind exploitation
✔ Detection-focused
✔ Scope-friendly
---
### 2️⃣ Normalisation & Noise Reduction
Instead of showing messy tool output, the framework:
- Normalises data (domains, URLs, parameters)
- De-duplicates repeated findings
- Filters non-actionable noise
This ensures the tester focuses on **signal, not spam**.
Example:
Multiple URLs → unique parameterised endpoints
Multiple ports → parsed open services
Multiple subdomains → consolidated attack surface
---
### 3️⃣ Correlation Engine (Human-like Logic)
This is the **core brain before AI**.
The framework correlates findings across tools to build **context**, such as:
- Open web ports + subdomains + login hints
- Parameterised URLs + SQLMap/XSS signals
- Nikto misconfigs + Nuclei template hits
- Many subdomains → higher chance of forgotten assets
Isolated issues are converted into **meaningful attack hypotheses**.
> Example logic:
Parameterised URL
SQLMap heuristic signal
XSStrike reflection
= High-value input validation hotspot
---
### 4️⃣ AI Reasoning Layer (Local LLM – Optional)
The AI layer uses **llama.cpp with GGUF models**, running **fully locally**.
The AI:
- Reads the correlated attack surface summary
- Mimics attacker-style reasoning
- Suggests **high-level attack paths**
- Explains **why certain areas matter more**
✔ Fully offline
✔ No cloud API
✔ Privacy-first
✔ Exploit-less (planning only)
AI is used for **decision support**, not automated hacking.
---
### 5️⃣ Attack Path Generation
Based on correlation + AI reasoning, the framework generates:
- Step-by-step **attack paths**
- Logical phases:
Recon → Entry Point → Expansion → Impact
- Priority scoring (Critical / High / Medium / Low)
- Risk context for each hotspot
This helps answer the real question:
> *“If I were attacking this legally, where would I start?”*
---
### 6️⃣ OWASP Top 10 Mapping (Signal-Based)
Each finding is approximately mapped to **OWASP Top 10 categories**, such as:
- A01 – Broken Access Control
- A03 – Injection (SQLi / XSS / Input issues)
- A05 – Security Misconfiguration
- A07 – Identification & Authentication Failures
⚠️ This is **signal-based mapping**, not a final verdict.
It is meant to make results:
- Report-ready
- Management-friendly
- Easier to explain to non-technical stakeholders
---
### 7️⃣ Risk Scoring & Prioritisation
The framework calculates a **rough risk score (0–100)** based on:
- Number of open ports
- Severity of Nuclei findings
- Correlated vulnerability signals
- Breadth of attack surface
This score is:
- Visual
- Educational
- Trendable (per session)
It is **not a CVSS replacement**, but a prioritisation aid.
---
### 8️⃣ Visualisation Layer (Streamlit UI)
All insights are presented through a clean Streamlit dashboard:
- Attack surface overview
- Risk metrics & trends
- Open port & subdomain graphs
- OWASP Top 10 tables
- Timeline view (Recon → Attack planning)
- Graphviz attack surface map
- High-level attack tree visualisation
No messy terminal output.
Only **clear offensive insight**.
---
### 9️⃣ Learning Mode & Explainability
Every major tool output can be:
- Explained in **beginner-friendly Hinglish**
- Interpreted using local AI (optional)
- Used as a learning reference
This makes the framework ideal for:
- Students
- Junior pentesters
- Interview preparation
- Red team mindset training
---
## ✨ Key Features (Expanded)
- 🔍 Multi-tool recon aggregation
- 🧠 Human-like vulnerability correlation
- 🤖 Local AI reasoning (llama.cpp, GGUF)
- 📊 Risk-based prioritisation & scoring
- 🧩 OWASP Top 10 signal mapping
- 🌐 Visual attack surface & attack tree
- 🧪 ParamSpider → SQLMap → XSStrike smart pipeline
- 📈 Risk trend tracking (session-based)
- 🧠 Explainable outputs (learning-first)
- ⚡ Fast Streamlit UI
- 🖥️ Offline / local-first architecture
- 📄 One-click professional reporting & export
- 🧾 Auto-generated attack surface & AI analysis report (Markdown)
- 🧠 Complete machine-readable scan bundle (JSON)
- ⏱️ Timestamped, domain-based filenames
- ⬇️ Instant download from the dashboard
- 🛡️ Legal, authorised & exploit-less by design
- 🎯 **Multiple Scan Modes**
- ⚙️ **Normal Mode** – Balanced recon & analysis for general security testing
- 🏆 **Bug Bounty Mode** – Low-noise, safe, scope-friendly scanning ideal for bounty programs
- 🎓 **Learning Mode** – Beginner-friendly explanations with AI-assisted reasoning (Hinglish support)
Each mode intelligently adjusts:
- 🧰 Tool execution behaviour
- 🔊 Noise vs signal balance
- 🧠 Explanation depth
This makes the framework usable for **both professionals and learners** 🚀
- 📦 **Smart Dependency Checker & Auto Installer**
- Automatically checks required & optional tools on startup
- Detects missing tools in the user’s system
- Prompts the user before installing anything
- Installs missing tools automatically (Windows – Chocolatey based)
- Skips tools that are already installed
- Ensures a smooth, beginner-friendly first-time setup
- No manual dependency hunting or broken PATH issues
---
## 🎥 Framework Demo Video
▶️ **Click below to watch the full demo**
(Shows real-time scanning, AI-assisted attack planning, visual graphs, and reporting flow)
[](https://github.com/HackerBlazeX/Offensive-AI-Attack-Path-Visualizer/releases)
---
## 📸 Framework Screenshots
### 1️⃣ Offensive Control Panel & AI Engine Selection
Centralized control panel with local AI engine selection, dependency checks, theme switching, and legal-only usage guardrails.
---
### 2️⃣ Tool Overview & Getting Started
Clean landing interface showing workflow guidance, scan modes, and offensive-security learning focus.
---
### 3️⃣ Scan Summary & Risk Snapshot
Auto-generated scan summary including resolved IP, scan timestamp, and overall risk score.
---
### 4️⃣Owasp Top 10 Coverage
Signal-based mapping of potential OWASP Top 10 risk categories derived from reconnaissance and scan outputs for prioritised review.
---
### 5️⃣ Timeline View-Recon to Attack Path Planning
Step-by-step visual flow from initial reconnaissance to structured attack-path planning for guided security testing.
---
### 6️⃣ Domain, IP, Ports & Subdomain Visual Graph
Session-level risk trend tracking to observe exposure changes during reconnaissance.
---
### 7️⃣ AI-Generated Attack Path & Ranking
High-level, AI-assisted attack path with asset ranking and testing priority to support informed and ethical security assessment planning.
---
### 8️⃣ Attack Tree & A.i Correlated Findings (Visual Graph)
Visual attack tree correlating reconnaissance findings, assets, and services to illustrate potential attack paths and relationships.
---
### 9️⃣ AI Risk Ranking
AI-based ranking of assets and findings based on exposure signals and contextual risk indicators.
---
### 🔟 Exploit Hint (AI-Assisted)
AI-assisted indicators to guide focused analysis and manual testing.
---
## ⚠️ Important Disclaimer
This framework is designed for:
- Legal & authorised security testing
- Education & learning
- Attack surface analysis
- Decision support
It does **NOT** provide:
- Exploit payloads
- Malware
- Illegal automation
Always follow scope, permissions, and local laws.
ℹ️ The framework never installs tools without explicit user consent.
---
## 🧰 Requirements
- Windows 10 / 11
- Python **3.10+**
- Git
- Streamlit
- llama.cpp (local LLM server)
---
## ⚙️ Installation (Windows – Easy)
```powershell
# 1️⃣ Clone the repository
git clone https://github.com/HackerBlazeX/Offensive-AI-Attack-Path-Visualizer.git
cd Offensive-AI-Attack-Path-Visualizer
# 2️⃣ Install dependencies
pip install -r requirements.txt
# 3️⃣ Start local LLM server (llama.cpp)
.\llama-server.exe -m path\to\model.gguf -c 4096 -t 6 -ngl 35
# 4️⃣ Run the framework
streamlit run Offensive-AI.py
# 5️⃣ Open in browser
http://localhost:8501
## ⚠️ Important: Hardcoded Paths Notice
Some paths inside the framework (for example **ParamSpider results directory, Nikto path, local tool locations**)
are currently **configured based on the developer’s local Windows environment**.
🔧 **What you need to do:**
- Review variables like:
- `PARAMSPIDER_BASE`
- `PARAMSPIDER_RESULTS_DIR`
- `nikto_path`
- Update them **according to your own system paths** if required.
💡 This design choice was made to:
- Keep the framework simple and readable
- Allow beginners to understand how tools interact
- Avoid complex environment abstractions in early versions
Future versions may introduce:
- Auto path detection
- Config file–based path management
✔️ Once paths are adjusted, the framework works normally.
⚠️ Important Note
This framework is not an auto-exploitation tool.
It is an AI-assisted offensive decision-support system designed to:
Reduce manual analysis time
Improve attack planning
Enhance learning and reporting quality
🔐 Legal Disclaimer
This project is intended only for authorised security testing, research, and education.
The author is not responsible for misuse or illegal activity.
📄 License
Licensed under the MIT License.
See the LICENSE file for details.
👨💻 Author
Dip Kar
Cybersecurity | Offensive Security | AI × Security
⭐ Support
If you find this project useful:
⭐ Star the repository
🧠 Share feedback
🚀 Contribute ideas or improvements