A production-grade, AI-driven Security Operations (SecOps) platform that autonomously detects, triages, and mitigates network threats. By leveraging multi-agent LLM orchestration and GitOps infrastructure-as-code, this system transforms raw network telemetry into verified, sandboxed, and auditable security patches in seconds.
- Web Command Center (Vercel): Autonomoussoc.vercel.app
- Automated GitOps Target Repo: DEMOCODE675/security-target-system
This architecture bridges advanced natural language synthesis with rigorous system administration and network packet analysis to ensure zero-trust execution.
- Real-Time Threat Triage: Ingests standard system logs (e.g., SSH brute-force, SYN floods, Nmap packet reconnaissance) and mathematically scores threat severity.
- Autonomous Code Synthesis: Generates targeted defensive Python scripts (e.g., dynamic
iptablesupdates) via Google Gemini 3.6 Flash without blocking delays. - Zero-Trust Execution: Runs synthesized mitigations inside ephemeral Python
subprocesssandboxes with strictly isolated resource allocation to prevent Remote Code Execution (RCE). - Human-in-the-Loop (HITL) Gate: Enforces security governance by halting execution for manual operator authorization before deployment.
- GitOps Compliance: Automatically stages successful mitigations to an external VCS repository, leaving a complete, auditable paper trail.
- Evaluation Pipeline: Generates automated
.jsonlsystem logs for continuous model benchmarking and future Direct Preference Optimization (DPO).
1. Clone the Repository
git clone https://github.com/DEMOCODE675/Autonomous-SOC.git
cd Autonomous-SOC
2. Create a .env file in the root directory
GEMINI_API_KEY=your_google_ai_key
GITHUB_TOKEN=your_personal_access_token
GITHUB_REPO=DEMOCODE675/security-target-system # or #( you can add your own target repo )
3. Launch the Intelligence Engine (Backend)
cd backend
pip install -r requirements.txt
python -m uvicorn main:app --reload --port 8000
4. Launch the Command Center (Frontend)
cd frontend
npm install
npm run dev