Rakshastra Documentation

Autonomous Cyber Defense Agent — Core Documentation

[ SYSTEM WARNING: CONFIDENTIAL OPERATIONAL MANUAL ] This manual outlines the architectural boundaries and configuration guidelines for the Rakshastra (रक्षास्त्र) OSINT defense daemon. Authorized operators must comply with law enforcement coordination directives.

1. System Architecture

Rakshastra is built on a modular three-tier autonomous agent loop designed for high-throughput social crawler ingestion:

Hermes Ingestion Core: Manages stealth head-free browser loops, rotating proxies, and direct API listeners targeting Telegram channels, WhatsApp groups, and public Instagram feeds.
Gemini Pro LLM Engine: Conducts advanced NLP parsing, decrypting emojis and slang (e.g. mapping ❄️ to active illicit substance keywords) and evaluating semantic threat weights.
Neo4j Identity Graph: Connects disparate handles, location coordinates, phone numbers, and cryptocurrency wallets into unified, traceable threat actors.

2. Getting Started

Deploying a local instance of the Rakshastra agent requires a modern Node.js environment and active endpoints for graph processing.

Install Dependencies

$ npm install @google/generative-ai neo4j-driver twilio dotenv

Run local crawler daemon

$ npm run agent -- --stream --visualize

3. Environment Configuration

System parameters are read from a standard .env file in the application root directory. Essential keys include:

# Gemini API Configuration
GEMINI_API_KEY=your_gemini_api_key_here

# Neo4j Graph Database
NEO4J_URI=bolt://localhost:7687
NEO4J_USER=neo4j
NEO4J_PASSWORD=your_secure_password

# Alert Dispatcher (Twilio)
TWILIO_ACCOUNT_SID=your_twilio_sid
TWILIO_AUTH_TOKEN=your_twilio_token
TWILIO_WHATSAPP_NUMBER=whatsapp:+14155238886

4. Stealth & Footprint Mitigation

To maintain anonymity while scraping illicit bot networks, Rakshastra routes active HTTP crawlers through a localized Tor control circuit.

User agent headers are rotated every 15 minutes, mimicking mainstream mobile devices, and active requests employ jitter delays (100ms - 2500ms) to bypass standard anti-scraping triggers.

Need Operational Support?

For setup assistance, custom regex rule contributions, or to report false positives, please log an issue on the repository page or consult the project developer channels.