franciscoderek7/omniaguard

# OmniaGuard **Autonomous Multi-Agent Cybersecurity Platform** ## Architecture OmniaGuard uses a **Multi-Agent Consensus Protocol** — no single agent decides. Every critical action requires cross-verification from multiple agents before execution. | # | Agent | Function | Catches | |---|---|---|---| | 1 | Network Scanner | Port/service discovery | Hidden backdoors | | 2 | Vulnerability Assessor | CVE matching | Known exploits | | 3 | Threat Intel | OSINT feeds | Emerging attacks | | 4 | Log Analyzer | SIEM parsing | Anomaly patterns | | 5 | Incident Responder | Automated containment | Breach escalation | | 6 | Phishing Detector | Email/URL analysis | Social engineering | | 7 | Malware Analyzer | Static/dynamic analysis | Zero-day malware | | 8 | Compliance Auditor | SOC2/ISO checks | Regulatory gaps | | 9 | Access Controller | IAM enforcement | Privilege abuse | | 10 | Data Loss Preventer | DLP monitoring | Insider threats | | 11 | Endpoint Protector | EDR coverage | Device compromise | | 12 | Cloud Guardian | AWS/GCP/Azure | Misconfiguration | | 13 | Dark Web Monitor | Credential leaks | Stolen data | | 14 | Report Generator | Executive summaries | Human-readable intel | ## Core Security Principles ### Zero-Trust Authorization Engine Every agent action must prove its legitimacy. No implicit trust. No inherited permissions. - Per-action authorization verification - Time-bounded permission grants - Automatic prompt injection detection - Failed attempts logged and trigger alerts ### Multi-Agent Consensus Protocol - Each agent votes on threat severity independently - Critical/high classifications require 3+ agent agreement - Dissenting opinions logged for audit trail - Weighted confidence scoring by domain expertise ### Autonomous 24/7 Operation | Time Block | Activity | |---|---| | 00:00–06:00 | Dark web scanning, threat intel updates | | 06:00–12:00 | Log analysis, anomaly detection | | 12:00–18:00 | Compliance checks, access reviews | | 18:00–00:00 | Vulnerability scans, report generation | Morning briefing via Telegram: what happened overnight, what needs your eyes. ## Tech Stack | Component | Technology | |---|---| | AI Inference | Together AI (Llama 3.1 70B) | | Database | Supabase (PostgreSQL) | | Bot Interface | Telegram Bot API | | Deployment | Docker / Supabase Edge Functions | | Language | Python 3.11+ | | DNS | Porkbun (omniaguard.com) | ## Quick Start git clone https://github.com/franciscoderek7/omniaguard.git cd omniaguard pip install -r requirements.txt cp .env.example .env # Edit .env with your keys python main.py --test # Verify all 14 agents python main.py # Start Telegram bot ## Deployment Options | Option | Cost | Best For | |---|---|---| | Local (polling) | $0 | Development, testing | | Supabase Edge Functions | $0 | Production serverless | | Docker on VPS | $5-10/mo | High-volume, dedicated | See `docs/deployment.md` for full instructions. ## Telegram Commands | Command | Function | |---|---| | `/start` | Initialize bot | | `/scan ` | Full security scan | | `/status` | System health check | | `/alerts` | View active alerts | | `/report` | Generate executive summary | | `/breach ` | Check breach databases | | `/monitor ` | Add to continuous monitoring | ## SR&ED Eligible This project qualifies for Canadian SR&ED tax incentives: - Multi-agent consensus algorithms (experimental development) - Zero-trust authorization for AI agents (systematic investigation) - Prompt injection defense pipelines (applied research) - Autonomous threat classification (technological uncertainty) All commits prefixed with `R&D:` for audit trail compliance. ## License MIT License ## Contact - **Platform:** OmniaGuard - **Organization:** Francisco Holdings - **GitHub:** @franciscoderek7

标签：后端开发