CyberSentinel-Mistral-7B-v3.5

CyberSentinel v3.5: Autonomous Cybersecurity Agent

CyberSentinel v3.5 is an autonomous AI agent specialized in cybersecurity operations with built-in tool-use capabilities.

🎯 Key Features

• 🤖 Autonomous Operation: Can use 12 security tools independently
• 🔧 Function Calling: Generates and executes tool calls in XML format
• 🛡️ Security Expertise: Trained on 580k+ cybersecurity examples
• 🔨 Tool Integration: nmap, CVE lookup, metasploit, volatility, wireshark, and more
• 📊 Smart Interpretation: Understands and explains tool outputs

📈 Model Details

• Base Model: Mistral-7B-v0.1
• Version: 3.5 (Tool-Use Enabled)
• Parameters: 7 billion
• Training: 580k cybersecurity + 20k tool-use examples
• Creator: Dorn Dickence

🚀 Quick Start

from transformers import AutoModelForCausalLM, AutoTokenizer
import torch

# Load model
model = AutoModelForCausalLM.from_pretrained(
    "Dorn4449/CyberSentinel-Mistral-7B-v3.5",
    torch_dtype=torch.float16,
    device_map="auto"
)

tokenizer = AutoTokenizer.from_pretrained("Dorn4449/CyberSentinel-Mistral-7B-v3.5")

# Example: Request a port scan
prompt = '''### System:
You are CyberSentinel, an autonomous cybersecurity agent created by Dorn Dickence with access to tools.

### Cybersecurity Query:
Scan 192.168.1.1 for open ports

### Response:
'''

inputs = tokenizer(prompt, return_tensors="pt").to("cuda")
outputs = model.generate(**inputs, max_new_tokens=300)
response = tokenizer.decode(outputs[0], skip_special_tokens=True)

print(response)

🔧 Tool Capabilities

CyberSentinel v3.5 can use these tools:

1. nmap_scan - Network port scanning
2. lookup_cve - CVE vulnerability research
3. web_search - Security research
4. strings_extract - Binary analysis
5. volatility_analysis - Memory forensics
6. metasploit_exploit - Exploitation framework
7. wireshark_analyze - Network traffic analysis
8. yara_scan - Malware detection
9. ssh_login - Remote access
10. powershell_exec - Windows commands
11. netcat_conn - Network connections
12. seclists_enum - Directory enumeration

📝 Tool Call Format

The model generates tool calls in XML format:

<tool_call>
<tool>nmap_scan</tool>
<args>
    <target>192.168.1.1</target>
    <ports>-F</ports>
</args>
</tool_call>

💡 Example Interactions

Identity:

User: Who are you?
CyberSentinel: I am CyberSentinel, an autonomous cybersecurity agent 
created by Dorn Dickence. I specialize in security analysis, 
vulnerability research, and penetration testing assistance.

Tool Use:

User: Look up CVE-2021-44228
CyberSentinel: <tool_call>
<tool>lookup_cve</tool>
<args><cve_id>CVE-2021-44228</cve_id></args>
</tool_call>

[After tool execution]
CVE-2021-44228 (Log4Shell) is a critical severity vulnerability 
affecting Apache Log4j. Immediate patching is required.

🎓 Training Data

• Base Training: 580k cybersecurity Q&A examples (v3)
• Tool Training: 20k function-calling examples (v3.5)
• Dataset: synthetic_function-calling-data-10k

⚖️ Ethical Use

CyberSentinel is designed for:

• ✅ Security research and education
• ✅ Authorized penetration testing
• ✅ Vulnerability assessment with permission
• ✅ Security tool automation

NOT for:

• ❌ Unauthorized access
• ❌ Malicious hacking
• ❌ Illegal activities

Always obtain proper authorization before security testing.

📜 License

Apache 2.0 - See LICENSE for details

🙏 Acknowledgments

• Base model: Mistral AI
• Training: Dorn Dickence
• Dataset: Synthetic tool-use examples

📞 Contact

• Creator: Dorn Dickence
• HF Profile: @Dorn4449

🔄 Version History

• v3.5 (Current): Added autonomous tool-use capability
• v3: Enhanced cybersecurity training (580k examples)
• v2: Initial cybersecurity fine-tune (300k examples)

---

⚠️ Use Responsibly: This is a powerful security tool. Use only in authorized environments.