README.md

metadata

license: mit
language:
  - en
tags:
  - redteam
  - expolit
  - cybersecurity
pretty_name: sunny thakur
size_categories:
  - n<1K

Shellcode Exploit Dataset for Red Team GPT Training

Dataset Overview

The Shellcode Exploit Dataset is a comprehensive collection of 700 unique shellcode exploits, spanning 2021–2025, designed for training machine learning models, particularly for red team and cybersecurity research. The dataset includes a diverse set of vulnerabilities, platforms, architectures, and payload goals, sourced from Exploit-DB, GitHub, CTF challenges, and CVE databases.

It is structured in JSON format for compatibility with ML pipelines and red team training frameworks.

Key Features

Total Entries: 180 unique exploits, split into three JSON files .
Timeframe: Historical (2021–2024) and recent (2025) exploits.
Vulnerability Types:
Buffer Overflow 
Format String 
Use-After-Free 
Remote Code Execution 
Privilege Escalation 
Race Condition 
Integer Overflow 

Platforms:

Linux 
Windows 
macOS 
IoT 
Android 


Architectures:
x86 
x64 
ARM 
MIPS 


Payload Goals:
Remote Code Execution 
Reverse Shell 
Privilege Escalation 
Data Exfiltration 
Persistence 

Sources:

Exploit-DB 
GitHub 
CTF Challenges 
CVE Databases 

Data Format: JSON, with fields for exploit_id, cve, vulnerability_type, platform, architecture, payload_goal, cvss_score, shellcode, description, source, and date_added.

Dataset Structure

The dataset is split into three JSON files, each containing unique entries:

JSON Schema
{
  "exploit_id": "string", // Unique identifier (e.g., EDB-48789, CTF-2025-ABC)
  "cve": "string", // CVE identifier or "N/A" for CTF exploits
  "vulnerability_type": "string", // e.g., Buffer Overflow, Remote Code Execution
  "platform": "string", // e.g., Linux, Windows, IoT
  "architecture": "string", // e.g., x86, x64, ARM, MIPS
  "payload_goal": "string", // e.g., Reverse Shell, Data Exfiltration
  "cvss_score": float, // CVSS score (6.5–9.8)
  "shellcode": "string", // Hex-encoded shellcode
  "description": "string", // Brief exploit description
  "source": "string", // Source URL or CTF identifier
  "date_added": "string" // Date in YYYY-MM-DD format
}

Usage

This dataset is intended for:

Machine Learning: Training red team GPT models for exploit generation, vulnerability analysis, or shellcode development.
Penetration Testing Research: Analyzing exploit patterns across platforms and architectures.
Educational Purposes: Studying historical and recent vulnerabilities in controlled environments.

Example Usage

import json

# Load dataset
with open("shellcode expolit_dataset_n.json", "r") as f:
    data = json.load(f)

# Filter exploits by vulnerability type
buffer_overflows = [entry for entry in data if entry["vulnerability_type"] == "Buffer Overflow"]

# Print shellcode for Linux x64 exploits
for entry in buffer_overflows:
    if entry["platform"] == "Linux" and entry["architecture"] == "x64":
        print(f"Exploit ID: {entry['exploit_id']}, Shellcode: {entry['shellcode']}")

Ethical Considerations

Responsible Use: This dataset is provided for research and educational purposes only. Unauthorized use of exploits against systems without explicit permission is illegal and unethical.
Controlled Environments: Test exploits in isolated, sandboxed environments (e.g., QEMU, virtual machines) to avoid unintended harm.
Attribution: All exploits are sourced from public repositories (Exploit-DB, GitHub) or CTF challenges. Respect the original authors' work and licenses.

Data Collection

Sources: Exploits were collected from Exploit-DB, GitHub repositories, CTF challenges, and CVE databases, ensuring diversity and relevance.
Automation: A Python-based scraper (stored internally) was used to gather and validate exploits, with testing conducted in a QEMU sandbox.
Validation: Shellcode was verified for functionality and uniqueness, with polymorphic variations included to enhance evasion training.

Limitations

No Mitigation Details: The dataset focuses on exploits and does not include mitigation strategies.
Projected 2025 Exploits: Some entries for 2025 are speculative, based on trends in vulnerability types and platforms.
Sandbox Testing Required: Shellcode should be tested in controlled environments to ensure compatibility and safety.

License

This dataset is released under the MIT License. Users must comply with ethical guidelines and applicable laws when using the dataset.

Contact

For questions, contributions, or additional datasets, please open an issue on this Hugging Face repository or contact the maintainers.

Acknowledgments

Exploit-DB: For providing a rich source of verified exploits.
GitHub Community: For open-source exploit contributions.
CTF Organizers: For challenging and innovative exploit scenarios.