danielostrow
/

c2sentinel

@@ -1,441 +1,34 @@
 ---
 license: mit
-language:
-  - en
-library_name: pytorch
-tags:
-  - security
-  - cybersecurity
-  - network-security
-  - c2-detection
-  - beacon-detection
-  - threat-detection
-  - malware-detection
-  - logbert
-  - transformer
-  - safetensors
-pipeline_tag: other
 ---
-# C2Sentinel
-[![Downloads](https://img.shields.io/badge/dynamic/json?url=https://huggingface.co/api/models/danielostrow/c2sentinel&query=downloads&label=Downloads&color=blue)](https://huggingface.co/danielostrow/c2sentinel)
-[![License](https://img.shields.io/badge/License-MIT-green.svg)](https://opensource.org/licenses/MIT)
-[![Demo](https://img.shields.io/badge/Demo-Hugging%20Face%20Spaces-yellow)](https://huggingface.co/spaces/danielostrow/c2sentinel)
-A machine learning model for detecting Command and Control (C2) beacon communications in network traffic. Built on a fine-tuned [LogBERT](https://arxiv.org/abs/2103.04475) transformer architecture.
-**Author:** Daniel Ostrow
-**Website:** [neuralintellect.com](https://neuralintellect.com)
-**Release Date:** January 18, 2026
----
-## Base Model
-This model is fine-tuned from the LogBERT architecture for log anomaly detection.
-- **Paper:** [LogBERT: Log Anomaly Detection via BERT](https://arxiv.org/abs/2103.04475) (Guo, Yuan, Wu - IJCNN 2021)
-- **Original Implementation:** [github.com/HelenGuohx/logbert](https://github.com/HelenGuohx/logbert)
----
-## Overview
-C2Sentinel analyzes network connection patterns to identify C2 beacon activity. The model uses behavioral analysis rather than port-based filtering, enabling detection of C2 communications on any port. This approach catches C2 activity regardless of whether attackers use expected ports (4444) or attempt to blend in on common ports (443, 80, 53).
-### Capabilities
-- Detection of 34+ C2 framework behavioral patterns across all ports
-- Slow beacon detection (intervals from seconds to hours)
-- Legitimate traffic pattern recognition (SSH keepalive, health checks, database connections)
-- Optional context enrichment (process information, reputation scores, threat intelligence)
-- IP reconnaissance and IOC generation
-- Safetensors format for secure model loading
----
-## Installation
-```bash
-pip install torch numpy safetensors huggingface_hub
-```
----
 ## Usage
-### Loading from HuggingFace Hub
-```python
-from c2sentinel import C2Sentinel
-sentinel = C2Sentinel.from_pretrained('danielostrow/c2sentinel')
-```
-### Loading from Local Files
-```python
-from c2sentinel import C2Sentinel
-sentinel = C2Sentinel.load('c2_sentinel')
-```
-### Analyzing Connections
-```python
-connections = [
-    {
-        'timestamp': 1000000,
-        'dst_ip': '10.0.0.1',
-        'dst_port': 443,
-        'bytes_sent': 200,
-        'bytes_recv': 500
-    },
-    {
-        'timestamp': 1000060,
-        'dst_ip': '10.0.0.1',
-        'dst_port': 443,
-        'bytes_sent': 200,
-        'bytes_recv': 500
-    },
-]
-result = sentinel.analyze(connections)
-if result.is_c2:
-    print(f"C2 detected: {result.c2_type}")
-    print(f"Probability: {result.c2_probability}")
-else:
-    print("No C2 detected")
-```
----
-## Connection Record Format
-| Field | Type | Required | Description |
-|-------|------|----------|-------------|
-| `timestamp` | float | Yes | Unix timestamp |
-| `dst_ip` | str | Yes | Destination IP address |
-| `dst_port` | int | Yes | Destination port |
-| `bytes_sent` | int | Yes | Bytes sent |
-| `bytes_recv` | int | Yes | Bytes received |
-| `src_ip` | str | No | Source IP address |
-| `src_port` | int | No | Source port |
-| `protocol` | str | No | Protocol (tcp/udp) |
-| `duration` | float | No | Connection duration in seconds |
----
-## Analysis Options
-### Threshold
-```python
-# Default threshold (0.5)
-result = sentinel.analyze(connections)
-# Lower threshold for higher sensitivity
-result = sentinel.analyze(connections, threshold=0.3)
-# Higher threshold for higher precision
-result = sentinel.analyze(connections, threshold=0.7)
-# Strict mode enforces minimum 0.7 threshold
-result = sentinel.analyze(connections, strict_mode=True)
-```
-### Context
-```python
-from c2sentinel import ConnectionContext
-context = ConnectionContext(
-    process_name='sshd',
-    known_good=True,
-    ip_reputation=0.95,
-    dns_queries=['api.example.com']
-)
-result = sentinel.analyze(connections, context=context)
-```
-### Whitelist and Blacklist
-```python
-sentinel.add_whitelist(ips=['8.8.8.8'], domains=['google.com'])
-sentinel.add_blacklist(ips=['10.10.10.10'], domains=['malware.example'])
-```
----
-## Result Object
-The `AnalysisResult` object contains:
-| Attribute | Type | Description |
-|-----------|------|-------------|
-| `is_c2` | bool | True if C2 detected |
-| `c2_probability` | float | Probability score (0.0-1.0) |
-| `c2_type` | str | Detected C2 framework type |
-| `confidence` | float | Model confidence |
-| `detection_method` | str | Method used (signature/ml/context/whitelist) |
-| `immediate_detection` | bool | True if signature-based |
-| `risk_factors` | list | Factors supporting C2 classification |
-| `mitigating_factors` | list | Factors against C2 classification |
-| `matched_legitimate_pattern` | str | Matched legitimate pattern name |
-| `service_type` | str | Detected service type |
-| `recommendations` | list | Suggested actions |
----
-## Batch Analysis
-```python
-connection_groups = [
-    [conn1, conn2, conn3],
-    [conn4, conn5, conn6],
-]
-results = sentinel.analyze_batch(connection_groups)
-```
----
-## Log File Parsing
-```python
-with open('conn.log', 'r') as f:
-    log_lines = f.readlines()
-results = sentinel.analyze_logs(log_lines, group_by_dst=True)
-```
-Supported formats: JSON, Zeek conn.log, syslog
----
-## Reconnaissance
-### IP Analysis
-```python
-info = sentinel.recon.analyze_ip('104.16.132.229')
-# Returns: is_valid, is_private, is_cdn, cdn_provider, reverse_dns
-```
-### Pattern Analysis
-```python
-patterns = sentinel.recon.analyze_connection_patterns(connections)
-# Returns: timing stats, volume stats, behavioral indicators
-```
-### IOC Generation
-```python
-if result.is_c2:
-    iocs = sentinel.recon.generate_iocs(connections, result.to_dict())
-    # Returns: ips, ports, timing_signatures, behavioral_indicators
-```
----
-## Detection Methodology
-### C2 Indicators
-- Consistent beacon intervals (low timing variance)
-- Consistent packet sizes (low size variance)
-- Single persistent destination
-- Balanced request/response ratio
-### Signature Detection
-Immediate detection for high-confidence C2 ports with matching behavioral patterns:
-- Port 4444 (Metasploit default)
-- Port 5555 (Metasploit alternative)
-- Port 31337 (Sliver)
-- Port 40056 (Havoc)
-### Legitimate Traffic Indicators
-- High response size variance
-- Asymmetric traffic patterns (small requests, large responses)
-- Multiple destinations
-- SSH keepalive patterns (small symmetric packets on port 22)
-- Health check patterns (regular intervals, variable response sizes)
----
-## Model Specifications
-| Specification | Value |
-|---------------|-------|
-| Architecture | LogBERT Transformer |
-| Parameters | 4.9 million |
-| Feature Dimensions | 40 |
-| Encoder Layers | 6 |
-| Attention Heads | 8 |
-| Hidden Dimension | 256 |
-| Format | Safetensors |
-| Size | 20 MB |
----
-## Training Your Own Model
-C2Sentinel supports training custom weights on your own data. This is useful for:
-- Fine-tuning on your network's specific traffic patterns
-- Adding detection for new C2 frameworks
-- Reducing false positives in your environment
-### Prerequisites
-```bash
-pip install torch numpy safetensors tqdm packaging
-```
-### Using Pre-trained Weights
-The released weights are trained on synthetic C2 beacon patterns covering 10+ framework types:
-```python
-from c2sentinel import C2Sentinel
-# Load pre-trained weights from HuggingFace
-sentinel = C2Sentinel.from_pretrained('danielostrow/c2sentinel')
-# Or load from local files
-sentinel = C2Sentinel.load('c2_sentinel')
-```
-### Training From Scratch
-Use the provided training script to train on synthetic data:
-```bash
-# Basic training (20,000 samples, 100 epochs)
-python train_model.py --epochs 100 --samples 20000
-# Faster training with fewer samples
-python train_model.py --epochs 50 --samples 10000
-# Custom learning rate
-python train_model.py --epochs 100 --samples 25000 --lr 0.0001
-```
-### Training on Custom Data
-Create a custom dataset class that returns connection records:
-```python
-from torch.utils.data import Dataset
-from c2sentinel import FeatureExtractor
-class CustomC2Dataset(Dataset):
-    def __init__(self, labeled_connections):
-        self.feature_extractor = FeatureExtractor()
-        self.samples = []
-        self.labels = []
-        for connections, is_c2 in labeled_connections:
-            features = self.feature_extractor.extract_features(connections)
-            self.samples.append(features)
-            self.labels.append(1 if is_c2 else 0)
-        # Normalize features (critical for training stability)
-        self.samples = np.array(self.samples, dtype=np.float32)
-        self.mean = np.mean(self.samples, axis=0)
-        self.std = np.std(self.samples, axis=0) + 1e-8
-        self.samples = (self.samples - self.mean) / self.std
-    def __len__(self):
-        return len(self.samples)
-    def __getitem__(self, idx):
-        return {
-            'features': torch.tensor(self.samples[idx]),
-            'label': torch.tensor(self.labels[idx], dtype=torch.float32)
-        }
-```
-### Fine-tuning Pre-trained Weights
-Start from pre-trained weights and fine-tune on your data:
-```python
-from c2sentinel import LogBERTC2Sentinel, C2SentinelConfig
-from safetensors.torch import load_file, save_file
-import torch.optim as optim
-# Load pre-trained model
-config = C2SentinelConfig()
-model = LogBERTC2Sentinel(config)
-state_dict = load_file('c2_sentinel.safetensors')
-model.load_state_dict(state_dict)
-# Fine-tune with lower learning rate
-optimizer = optim.AdamW(model.parameters(), lr=0.00005, weight_decay=0.01)
-# Train on your data...
-# Save fine-tuned weights
-save_file(model.state_dict(), 'c2_sentinel_finetuned.safetensors')
-```
-### Training Tips
-1. **Feature Normalization**: Always normalize input features. Save the mean/std for inference:
-   ```python
-   np.savez('normalization_params.npz', mean=mean, std=std)
-   ```
-2. **Learning Rate**: Use 0.0001 for training from scratch, 0.00005 for fine-tuning
-3. **Gradient Clipping**: Prevent exploding gradients:
-   ```python
-   torch.nn.utils.clip_grad_norm_(model.parameters(), max_norm=1.0)
-   ```
-4. **Early Stopping**: Monitor validation accuracy and stop when it plateaus
-5. **Balanced Data**: Use roughly equal C2 and benign samples
-### Model Output Files
-After training, you'll have:
-- `c2_sentinel.safetensors` - Model weights
-- `normalization_params.npz` - Feature normalization parameters
-- `c2_sentinel.json` - Model configuration
----
-## Files
-```
-c2sentinel/
-    c2sentinel.py              # Main module
-    c2_sentinel.safetensors    # Model weights
-    c2_sentinel.json           # Model configuration
-    README.md                  # Documentation
-    API_REFERENCE.md           # API reference
-    examples/
-        basic_usage.py
-        advanced_usage.py
-```
----
-## License
-MIT License
-Copyright (c) 2026 Daniel Ostrow
-Permission is hereby granted, free of charge, to any person obtaining a copy of this software and associated documentation files (the "Software"), to deal in the Software without restriction, including without limitation the rights to use, copy, modify, merge, publish, distribute, sublicense, and/or sell copies of the Software, and to permit persons to whom the Software is furnished to do so, subject to the following conditions:
-The above copyright notice and this permission notice shall be included in all copies or substantial portions of the Software.
-THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.

 ---
+title: C2Sentinel
+emoji: 🛡️
+colorFrom: red
+colorTo: gray
+sdk: docker
+pinned: false
 license: mit
+models:
+  - danielostrow/c2sentinel
 ---
+# C2Sentinel Demo
+Interactive demo for C2Sentinel - a machine learning model for detecting Command and Control (C2) beacon communications in network traffic.
+## Features
+- Analyze network connection patterns for C2 activity
+- Preset examples for common scenarios (C2 beacons, legitimate traffic)
+- Adjustable detection threshold
+- Detailed risk factor analysis
 ## Usage
+1. Paste connection data as JSON or select a preset example
+2. Adjust the detection threshold if needed
+3. Click "Analyze" to run the model
+## Model
+See the [C2Sentinel model repository](https://huggingface.co/danielostrow/c2sentinel) for full documentation.
+**Author:** Daniel Ostrow | [neuralintellect.com](https://neuralintellect.com)