🛡️ OpenSentinel

AI-Powered SOC Command Center

Natural Language → SIEM Queries · Real-Time Detection · AI Copilot · Automated Response

Features • Architecture • Quickstart • API • Security • Contributing

---

Overview

OpenSentinel is a full-stack Security Operations Center (SOC) platform that combines AI-driven analysis with traditional SIEM capabilities. It translates natural language queries into Splunk SPL, runs automated detection rules mapped to MITRE ATT&CK, and provides an AI copilot for incident investigation — all behind an API hardened against the OWASP API Security Top 10.

Features

🤖 NLP Query Engine

• Natural language → Splunk SPL translation via spaCy NER
• Support for custom fine-tuned NLP models
• Entity extraction: IPs, domains, users, security actions
• Context-aware query building with time range support

🔍 Sentinel Detection Engine

• 6 built-in detection rules mapped to MITRE ATT&CK
• Real-time alert lifecycle (new → acknowledged → closed)
• Configurable thresholds and time windows
• Anomaly detection module

Rule	MITRE	Severity
Brute Force Detection	T1110	High
Suspicious PowerShell	T1059.001	Critical
DNS Exfiltration	T1048.003	Critical
Lateral Movement (SMB)	T1021.002	High
Privilege Escalation	T1068	Critical
C2 Beaconing	T1071	High

📋 Incident Management

• Full incident lifecycle with timeline tracking
• Alert-to-incident linking
• Notes, status updates, and severity tracking
• BOLA-protected object-level access control

🧠 AI Copilot (Google Gemini)

• Script Analyzer — Deobfuscate and assess suspicious scripts
• Threat Intel Enrichment — IOC lookups via VirusTotal & AbuseIPDB
• Investigator — AI-guided investigation workflows
• Summarizer — Natural language summaries of SIEM query results

🎯 Threat Hunting

• Pre-built hunt library with categorized queries
• Hunt notebook for tracking investigations
• One-click hunt execution against live SIEM data

⚡ Automated Playbooks

• 5 built-in response playbooks
• Action-based execution engine
• Execution statistics and audit trail

🖥️ Dashboard

• Glassmorphism dark-theme UI
• Responsive design with hamburger menu
• Real-time status indicators
• Interactive chat interface for NLP queries

Architecture

┌─────────────────────────────────────────────────────┐
│ Dashboard (HTML) │
├─────────────────────────────────────────────────────┤
│ FastAPI Server (main.py) │
│ ┌──────────┐ ┌──────────┐ ┌───────────────────┐ │
│ │ Security │ │ Auth │ │ Input Validator │ │
│ │Middleware│ │ Manager │ │ (SSRF/Injection) │ │
│ └──────────┘ └──────────┘ └───────────────────┘ │
├─────────────────────────────────────────────────────┤
│ ┌──────────┐ ┌──────────┐ ┌──────────┐ │
│ │ SOC │ │ Sentinel │ │ AI │ │
│ │ Agent │ │ Engine │ │ Copilot │ │
│ │(NLP→SPL)│ │(Detect) │ │(Gemini) │ │
│ └────┬─────┘ └──────────┘ └────┬─────┘ │
│ │ │ │
│ ┌────┴─────┐ ┌──────────┐ ┌────┴─────┐ │
│ │ Splunk │ │Playbooks │ │VirusTotal│ │
│ │ SIEM │ │ Engine │ │AbuseIPDB │ │
│ └──────────┘ └──────────┘ └──────────┘ │
└─────────────────────────────────────────────────────┘

Quickstart

Prerequisites

• Python 3.10+
• Splunk instance (optional — Gemini AI works without SIEM)
• API Keys for Gemini, VirusTotal, AbuseIPDB (optional)

Installation

# Clone the repository
git clone https://github.com/Techris93/OpenSentinel.git
cd OpenSentinel
# Create virtual environment
python -m venv venv
source venv/bin/activate # macOS/Linux# venv\Scripts\activate # Windows# Install dependencies
pip install -r requirements.txt
# Download spaCy model
python -m spacy download en_core_web_sm

Configuration

Create a .env file in the project root:

# Google Gemini API (https://aistudio.google.com/app/apikey)GEMINI_API_KEY=your_gemini_key_here# VirusTotal API (https://www.virustotal.com)VIRUSTOTAL_API_KEY=your_vt_key_here# AbuseIPDB API (https://www.abuseipdb.com)ABUSEIPDB_API_KEY=your_abuseipdb_key_here# Splunk Connection (optional)SPLUNK_HOST=localhostSPLUNK_PORT=8089SPLUNK_USERNAME=adminSPLUNK_PASSWORD=changeme# SecurityOPENSENTINEL_API_KEY=your-secure-api-keyRATE_LIMIT=60CORS_ORIGINS=http://localhost:5001

Run

python main.py

🛡️ OpenSentinel Command Center v1.0.0 starting on port 5001...
API Docs: http://localhost:5001/api/docs
Dashboard: http://localhost:5001/dashboard
OWASP: API1-10 protections active

API Reference

All endpoints are available at /api/v1/ (versioned) and /api/ (legacy).

Public Endpoints

Method	Path	Description
GET	/api/v1/health	Health check with system status
GET	/dashboard	Web dashboard UI
GET	/api/docs	Interactive OpenAPI docs

Authenticated Endpoints

Include X-API-Key: <your-key> header in all requests.

Method	Path	Role	Description
POST	/api/v1/connect	admin	Connect to Splunk SIEM
POST	/api/v1/chat	analyst	NLP query interface
POST	/api/v1/sentinel/start	admin	Start detection engine
POST	/api/v1/sentinel/stop	admin	Stop detection engine
GET	/api/v1/sentinel/alerts	analyst	Get security alerts
GET	/api/v1/anomalies	analyst	Get detected anomalies
GET	/api/v1/incidents	analyst	List incidents (BOLA-filtered)
GET	/api/v1/incidents/{id}	analyst	Get incident detail
GET	/api/v1/hunts	analyst	List threat hunts
POST	/api/v1/hunts/{id}/execute	analyst	Execute a threat hunt
POST	/api/v1/copilot/analyze-script	analyst	AI script analysis
POST	/api/v1/copilot/enrich	analyst	IOC enrichment
GET	/api/v1/playbooks	analyst	List playbooks
GET	/api/v1/playbooks/stats	analyst	Playbook execution stats

Example: Chat Query

curl -X POST http://localhost:5001/api/v1/chat \
-H "X-API-Key: your-key" \
-H "Content-Type: application/json" \
-d '{"message": "Show me failed login attempts from 192.168.1.100"}'

Security

OpenSentinel implements all OWASP API Security Top 10 protections:

#	Vulnerability	Mitigation
API1	BOLA	Ownership verification on incident endpoints
API2	Broken Auth	Key expiry, timing-safe comparison, brute-force lockout
API3	Property Auth	Field-level response filtering by role
API4	Rate Limiting	Global + per-endpoint rate limits
API5	Function Auth	Role-based access (admin/analyst/readonly)
API6	Anti-Automation	Per-endpoint throttling on sensitive flows
API7	SSRF	Private IP blocking in recon targets
API8	Misconfiguration	Security headers (HSTS, CSP, X-Frame-Options)
API9	Inventory	API versioning + OpenAPI documentation
API10	Unsafe Consumption	External API response truncation and sanitization

Role Hierarchy

admin > analyst > readonly

• admin — Full access: SIEM connection, sentinel control, all data
• analyst — Query, investigate, hunt, view own incidents
• readonly — View limited incident fields only

Project Structure

OpenSentinel/
├── main.py # FastAPI server (19 endpoints)
├── soc_agent.py # NLP → Splunk SPL engine
├── gemini_client.py # Google Gemini AI client
├── dashboard.html # Web UI (glassmorphism)
├── requirements.txt # Python dependencies
├── .gitignore
│
├── security/
│ ├── __init__.py
│ ├── auth.py # API key management & RBAC
│ ├── middleware.py # Rate limiting, headers, audit
│ └── input_validator.py # Injection & SSRF prevention
│
├── sentinel/
│ ├── __init__.py
│ ├── detection_engine.py # Rule-based threat detection
│ ├── rules.py # MITRE ATT&CK detection rules
│ ├── anomaly_detector.py # Statistical anomaly detection
│ └── incidents.py # Incident lifecycle manager
│
├── copilot/
│ ├── __init__.py
│ ├── summarizer.py # AI event summarization
│ ├── script_analyzer.py # Malicious script analysis
│ ├── investigator.py # AI investigation assistant
│ └── threat_intel.py # VirusTotal & AbuseIPDB
│
├── hunting/
│ ├── __init__.py
│ ├── hunt_library.py # Pre-built hunt queries
│ └── notebook.py # Hunt investigation notebook
│
├── playbooks/
│ ├── __init__.py
│ ├── engine.py # Playbook execution engine
│ └── actions.py # Response action definitions
│
└── connectors/
└── __init__.py # SIEM connector interface

Contributing

Contributions are welcome! Areas where help is especially valued:

• Detection Rules — Add new MITRE ATT&CK-mapped rules
• SIEM Connectors — Extend beyond Splunk (Elastic, QRadar, Sentinel)
• Playbook Actions — Add automated response capabilities
• Tests — Improve test coverage
• Documentation — Improve guides and tutorials

License

MIT License — see LICENSE for details.

---

Built with ❤️ for the Blue Team