a-vip/sovereign-intel-engine

# 📡 Sovereign Intelligence Engine An autonomous, high-fidelity threat intelligence (OSINT) harvester, geocoding system, and strategic weekly compiler. The engine scans RSS feeds and GDELT geopolitics signals, filters exclusively for high-priority AI governance, lethal autonomous weapons (LAWS), border surveillance, and state IHL compliance, geocodes coordinates via Gemini 3.5 Flash, and synchronizes the results seamlessly to the cloud database of the **Sovereign Dashboard**. ## ⚙️ Architecture Workflow graph TD A[RSS Feeds] --> D[intel_engine.py] B[GDELT doc API] --> D D --> E{AI Relevance Double-Filter} E -- Rejected --> F[Discard] E -- Match --> G[Geocoding Engine] G --> H[Gemini 3.5 Flash Model Loop] H --> I[Generate Obsidian Tactical Dossier] H --> J[Draft Social Outbound Broadcast] I --> K[Full Vault Sync Daemon] K --> L[Sovereign Dashboard Cloud Database] L --> M[Live Geo-Map Vis] ## 🛠️ Main Features * **Dual-Telemetry OSINT Harvester**: Combines high-resolution RSS feeds with deep GDELT geopolitics documentation data for both real-time and robust historical backfilling. * **Double-Filter Relevance Checker**: Incorporates a strict blocklist (blocking sports, celebrity gossip, and general trivia) and a targeted positive regex checker to capture exactly technology governance, drone systems, and biometric/border policing risks. * **Bulletproof Multi-Model Fallback**: Loops across multiple Gemini models (`gemini-3.5-flash`, `gemini-3.1-flash-lite`, `gemini-2.5-flash`, etc.) when geocoding and compiling briefs. If an individual model hits a `429` rate limit or a timeout, the engine seamlessly calls the next model to guarantee absolute execution stability. * **Automated Backfiller**: Command-line historical parameter overrides for fast retrospective backdating (e.g. `python intel_engine.py 2026-05-25`). * **Mastodon Integration**: Includes a human-in-the-loop verify-and-publish client to draft micro-broadcast alerts for advocacy campaigns. * **Obsidian Synchronization Daemon**: Syncs threat folders directly to Vercel hosted databases via an authenticated syncing API. ## 📁 Repository Structure * `intel_engine.py`: The heart of the automation. Scrapes RSS/GDELT, handles AI relevance and geocoding, generates Obsidian threat dossiers, drafts broadcasts, and handles backfills. * `vault_sync.py`: Synchronizes the Obsidian intelligence vault to the live Vercel-hosted Sovereign Dashboard database. * `broadcast_manager.py`: Human-in-the-loop Mastodon client to review, edit, and dispatch outbound OSINT broadcast drafts. * `intel_sources.json`: Centralized configuration directory containing tracked RSS feeds and relevance keywords. * `Run-OSINT-Crawler.ps1`: Powershell automation loader to spin up deeper targeted searches on selected defense and legal entities. * `intel_backdate.py`: Standalone historical test crawler. * `compile_weekly.py`: Standalone weekly briefing compiler. ## 🚀 Setup & Setup Instructions 1. **Clone the Repository**: git clone https://github.com/a-vip/sovereign-intel-engine.git cd sovereign-intel-engine 2. **Install Python Dependencies**: pip install -r requirements.txt 3. **Environment Setup (`.env`)**: Create a `.env` file in the root directory and configure the active API tokens: GEMINI_API_KEY=your_gemini_api_key DASHBOARD_API_TOKEN=your_dashboard_api_token MASTODON_API_TOKEN=your_mastodon_token 4. **Usage**: * **Tactical Daily Run**: python intel_engine.py * **Retrospective Backfill**: python intel_engine.py 2026-05-25 * **Broadcast Alerts Review**: python broadcast_manager.py ## ⚖️ License MIT License. Created for the Sovereign Algorithms Campaign. ## 👨‍💻 About the Creator This crawler and sync daemon suite was designed and built by **Avi Perera**. * 🌐 **Personal Website**: [aviperera.com](https://aviperera.com) * 🚀 **Enterprise Platform**: [sovdash.com](https://sovdash.com) (Sovereign Intel Platform)