Hugging Face's logo

PandemicDragon
/
CORTANA-AI

Model card Files
xet
 Community

🛰️ 24/7 Cortana & J.A.R.V.I.S. Cognitive Telemetry Construct

An autonomous tactical intelligence construct combining the strategic threat synthesis of Cortana (HALO) and the dry operational precision, multimodal voice/vision execution, and persistent memory of J.A.R.V.I.S. (Iron Man).

Runs 24/7 with zero VPS and zero paid domain requirements using Telegram Long Polling, Groq LLM/Whisper/Vision inference, Supabase PostgreSQL episodic memory, and a glassmorphic Vercel Vocal HUD.

🏛️ Cognitive System Architecture 

graph TD
    subgraph Operator Inputs
        Voice[🎙️ Telegram Voice Notes] --> Poller[bot.py Cognitive Engine]
        Photos[📸 Telegram Receipt / Ticket Photos] --> Poller
        Text[💬 Natural Language Directives] --> Poller
    end

    subgraph Sensory Telemetry (Keyless)
        Meteo[Open-Meteo Atmospheric Sensors] --> Poller
        CISA[CISA KEV Exploit Catalog] --> Poller
        Feeds[News & Threat Intelligence Feeds] --> Poller
    end

    subgraph Brain & Tool-Calling (ReAct Pattern)
        Poller <-->|llama-3.3-70b-versatile| Groq[Groq AI Cloud]
        Poller <-->|whisper-large-v3-turbo| Groq
        Poller <-->|Vision OCR| Groq
    end

    subgraph Persistent Memory & Operations
        Poller <-->|Operator Directives & Tech Stack| DB[(Supabase PostgreSQL)]
        Poller -->|Tasks & Deadlines| DB
        Poller -->|M-Pesa / Receipt Ledger| DB
        Poller -->|Voice-to-Action Vault| DB
        Poller -->|Tactical SitReps| DB
    end

    subgraph Real-Time Display & Vocal HUD
        DB -->|Public Read-Only anon RLS| Web[🌐 Vercel Web Dashboard<br/>index.html]
        Web -->|Web Speech API Vocalization| Speaker[🔊 Vocal SitRep Audio]
    end

🧠 Autonomous Core Capabilities

Capability Model / Engine Description
Tactical Persona llama-3.3-70b-versatile Cortana threat synthesis + J.A.R.V.I.S. operational precision. No fluff, no emojis.
Voice Directives whisper-large-v3-turbo Speak to your bot on Telegram; transcribes at 228x real-time speed and executes tool loops.
Vision OCR llama-3.2-11b-vision Send photos of paper receipts, invoices, or tickets to automatically parse M-Pesa / amount / dates.
Episodic Memory bot_memory table Stores standing orders and your tech stack, cross-referencing incoming CVEs against your environment.
Deadlines & Reminders check_due_tasks loop Automatically tracks deadlines and pushes alert pings to your phone when tasks are due.
Personal Ledger personal_ledger table Automatically organizes M-Pesa statements, merchant receipts, and income into an indexed ledger.
Voice Idea Vault idea_vault table Instant spoken logging for code snippets, song lyrics, or project architectures.
Vocal Web HUD window.speechSynthesis Web dashboard that reads newly synthesized SitRep briefings aloud with one click.
Wi-Fi Cyber Scanner Async Socket & Banner Fingerprint Dedicated multi-port scanner auditing 20-35 services on Wi-Fi devices (Google Cast, Roku, ADB, HTTP, SSH) across mobile and desktop.
Secure Local CLI & Workspace Bridge halo_agent.py (Gemini + Groq) Dual-listener daemon (Port 5050 + Supabase command_queue) for zero-latency local code refactoring, colorized unified diffs, path-sandboxed writes, and git stash + syntax rollback safety.
Autonomous Sub-Agent Swarm swarm_engine.py (Queen-Worker Architecture) Cortana dynamically spawns, commissions, and commands unlimited super-powerful specialized sub-agents (Security, Code Architecture, Quant Finance, OSINT, Music Promo, Frontier AI). All agents execute parallel workloads with AST validation and Hydra LLM failover, obeying Cortana with 100% unconditional fidelity.
Africa's Talking Shortcodes (SMS) api/sms.js (GSM 03.40 / SS7 Signaling) Chat with Halo over 2-way Africa's Talking Shortcodes (e.g. 20880 or sandbox 99999) across Safaricom and Airtel Kenya with zero mobile data bundles required.
Sovereign WhatsApp Gateway api/whatsapp.js (Meta Cloud & AT WhatsApp) 24/7 rich interactive WhatsApp connectivity supporting Meta Cloud WhatsApp API, Africa's Talking WhatsApp, and Twilio WhatsApp with real-time tactical sitreps, tool links, and Telegram cross-channel sync.
High-Tech Typography Matrix Google Fonts + CSS Engine Replaces cramped legacy fonts with movie-grade Oxanium display headers, Plus Jakarta Sans & Inter hyper-crisp body text, Space Grotesk HUD labels, JetBrains Mono telemetry, and an interactive 6-preset selector (Quantum, Stark, Cortana, Minimalist, Night City, Retro).

📁 Repository Structure 

MYBOT/
├── bot.py                    # 24/7 Cortana / J.A.R.V.I.S. cognitive daemon & tool executor
├── halo_agent.py             # Autonomous principal systems engineer & local workspace bridge (CLI + Port 5050)
├── wifi_manager.py           # 802.11 RF spectrum scanning and authenticated association engine
├── speak.py                  # Direct physical audio driver & terminal TTS bridge
├── laptop_agent.py           # Remote workstation hardware listener
├── core_bot.py               # Baseline custom web collector bot
├── test_cyber_telemetry.py   # Test utility for CISA KEV JSON & Threat feeds
├── test_weather_news.py      # Test utility for Open-Meteo & RSS feeds
├── schema.sql                # Supabase schema (Memory, SitReps, Tasks, Ledger, Vault, RLS)
├── .env.example              # Environment variables template (Telegram, Supabase, Groq)
├── vercel.json               # Vercel deployment configuration
├── index.html                # Cognitive Command HUD (Vocalizer + Ops Matrix + Telemetry)
├── style.css                 # Cyber-telemetry CSS design system
├── app.js                    # Frontend data engine, Web Speech API & settings modal
└── requirements.txt          # Python dependencies

⚡ Quick Setup & Launch

1. Database Setup (Supabase)

  1. Open your Supabase Dashboard -> SQL Editor -> New Query.
  2. Run the queries from schema.sql to provision all tables (bot_memory, tactical_briefings, personal_tasks, personal_ledger, idea_vault, weather_reports, news_articles, cyber_alerts) and public RLS policies.
  3. Under Project Settings -> API, copy:
    • Project URL (SUPABASE_URL)
    • service_role key (SUPABASE_KEY for Python bot)
    • anon key (SUPABASE_ANON_KEY for Vercel Web HUD)

2. Groq AI API Key (Free)

  1. Navigate to console.groq.com and create a free account.
  2. Generate an API Key -> GROQ_API_KEY.

3. Telegram Bot Setup

  1. Message @BotFather on Telegram and run /newbot to get your TELEGRAM_TOKEN.
  2. Message @userinfobot to get your numeric TELEGRAM_CHAT_ID.
  3. Send /start to your new bot once to initialize the session.

4. Configure Local .env

Copy .env.example to .env:

cp .env.example .env

Update .env with your credentials:

TELEGRAM_TOKEN=123456789:ABCdefGHIjklMNOpqrSTUvwxYZ
TELEGRAM_CHAT_ID=123456789
SUPABASE_URL=https://your-project.supabase.co
SUPABASE_KEY=your_supabase_service_role_secret
SUPABASE_ANON_KEY=your_supabase_anon_public_key
GROQ_API_KEY=gsk_your_groq_api_key_here
SYNC_INTERVAL=300
TELEGRAM_DISPATCH_INTERVAL=1800

.\.venv\Scripts\python setup_phone.py

This utility validates your Telegram credentials and sends an instant test ping to your phone.

6. Launch the 24/7 Autonomous Daemon

Run the cognitive engine:

.\.venv\Scripts\python bot.py

Your phone will receive an initialization ping:

Tactical Intelligence Construct Online. Kinetic War, Macro-Economic Shocks & Protective Protocols active.

💬 How to Interact With Cortana / J.A.R.V.I.S.

You don't need rigid commands anymore; talk or speak naturally via Telegram:

  • On-Demand Intelligence Updates ("when I ask"):
    • "Give me an update" / "SitRep" / "Briefing" / "Latest update" / "What's happening" -> Instantly returns the full, high-fidelity tactical intelligence briefing on demand, resetting the 30-minute broadcast timer.
  • Disciplined Broadcast Cadence:
    • Routine proactive SitReps to Telegram are strictly rate-limited to after 30 minutes (TELEGRAM_DISPATCH_INTERVAL=1800). Continuous sensor sweeps (every 5 mins) silently refresh Supabase and the live Web HUD without cluttering your phone.
  • Deep OSINT & Human Profiling: "Investigate Alex Vance—we have a meeting tomorrow regarding a software partnership."
    -> Sweeps LinkedIn, X, GitHub, Reddit, Kenya business filings & PDFs concurrently. Computes psychological drivers, cognitive biases, and gives 3 negotiation anchors.
  • Crowd Psychology & Sentiment Analysis: "Analyze crowd psychology right now around NSE banking stocks and interest rate sentiment."
    -> Scans Reddit, HN, X dorks, and financial media to diagnose FOMO vs. capitulation and compute contrarian positioning.
  • Geospatial Reconnaissance: "Run a deep dive on Nairobi Commercial Corridor and its strategic value."
    -> Queries OpenStreetMap Nominatim, local infrastructure data, and economic drivers.
  • Negotiation & Message Deconstruction: "Why is this person acting defensive in these messages? [Forwarded text]"
    -> Cross-references against behavioral psychology & negotiation doctrine in bot_memory, decoding status drivers and crafting frame-control responses.
  • War & Macro Query: "Give me the latest Red Sea war impact on Kenya shipping and how to position my NSE portfolio."
    -> Queries live conflict_and_macro_intel and synthesizes defensive & asymmetric financial maneuvers.
  • Voice Memo: Hold the mic and say: "Remind me to push the Docker containers tomorrow at 10 AM EAT."
    -> Transcribed via Whisper, calculates timestamp, schedules task in database, and notifies you when due.
  • Two-Way Voice Dialogue: Speak to the bot via voice note. Cortana will think, execute commands, reply in text, AND send a neural voice note (en-US-AriaNeural) straight back to your Telegram chat while optionally vocalizing through your room speakers.
  • Smart TV Control: "Turn on the Living Room TV and put YouTube on."
    -> Sends Wake-on-LAN packet, connects via wireless ADB (adb connect <ip>:5555), and launches the URL or powers on the screen.
  • Nearby Wi-Fi Reconnaissance & RF Spectrum Scan:
    • "Scan nearby wifis" / "View nearby wifis" / "Show available wifi networks" -> Performs an in-flight RF sweep across 2.4 GHz, 5 GHz and 6 GHz bands. Returns a formatted tactical matrix detailing visible SSIDs, signal percentages, signal bars, channels, radio types (802.11ax/ac/n), and security standards (WPA2-Personal/WPA3/Open).
  • Authenticated Wi-Fi Association:
    • "Connect to wifi Olalo-5G with password 2011@Olalo" or "Connect wifi "MyNetwork" "secretKey"" -> Dynamically generates a native Windows WLAN XML profile conforming to Microsoft networking schemas, associates the interface, verifies 4-way WPA handshake completion, and purges temporary credential artifacts.
    • "What wifi am I connected to?" / "Wifi status" -> Retrieves live adapter telemetry including connected SSID, BSSID, RSSI/Signal %, radio channel, and Rx/Tx data link rates.
    • "Disconnect wifi" -> Gracefully unbinds the active WLAN interface.
  • Wi-Fi Network Device Sweep: "Scan my Wi-Fi network and tell me what gadgets are connected."
    -> Sweeps the local ARP cache table, detects all connected TVs, laptops, and IoT nodes, and synchronizes their online status with Supabase.
  • Laptop Operational Control (Speech, Chrome & Hardware):
    • "Speak: Protocol activated and standing by." -> Vocalizes directly through your laptop's physical speakers via Windows SAPI / System TTS.
    • "Open Chrome" or "Open Chrome https://youtube.com" -> Launches Google Chrome browser or navigates directly to a target destination.
    • "Close Chrome" -> Immediately terminates and force-kills all running Chrome browser processes cleanly.
    • "Turn up laptop volume" / "Volume down" / "Mute laptop" -> Adjusts workstation master volume level.
    • "Lock laptop" -> Locks the Windows workstation screen instantly.
  • Laptop & PC Automation: "Wake up my laptop and open the tactical HUD."
    -> Transmits WoL magic packet to the PC's MAC address and launches the command.
  • Receipt / M-Pesa Photo: Send a photo of a restaurant receipt or M-Pesa screenshot.
    -> Vision model extracts merchant name, amount, date, reference code, and logs it to your ledger.
  • Tactical Query: "Cortana, are any active CVEs impacting PostgreSQL or Linux right now?"
    -> Executes query_telemetry tool call and responds with strategic impact assessment.
  • Web Inspection: "Check the status page at https://example.com and summarize any outages."
    -> Executes live_web_probe tool call and provides instant tactical synopsis.
  • Idea Logging: "Log a project idea: Decentralized voice mesh using libp2p."
    -> Executes log_vault_note and stores in your Idea Vault.

🎙️ Two-Way Natural Voice & Ambient Room Listener

1. Cortana Halo Persona & Neural TTS Voice

The construct uses Microsoft's Neural Speech engine (edge-tts) with zero API keys and zero subscriptions required:

  • Default Voice: en-US-AriaNeural (Halo Cortana persona: calm tactical authority, razor-sharp intellect, protective cadence).
  • Alternative Style: en-GB-RyanNeural (British J.A.R.V.I.S. cadence).
  • Every response generated by Cortana is synthesized into high-fidelity neural audio, pushed back as an instant playable voice note to Telegram (sendVoice), and played through host speakers when enabled.

2. Ambient Room Wake-Word Listener ("Cortana" / "Jarvis")

Run the bot with the --wake flag or set ENABLE_WAKEWORD=true in your .env:

.\.venv\Scripts\python bot.py --wake --speak

When Cortana hears "Cortana" or "Jarvis" spoken out loud in your room, it transcribes your speech via Groq Whisper (whisper-large-v3-turbo) in ~250ms, executes your directive, and answers aloud over your room/laptop speakers.

3. Direct System-Level Audio Driver & Terminal TTS Bridge (speak.py)

Direct COM hook into your laptop's physical sound card (via in-process win32com SAPI and edge-tts Cortana neural engine):

  1. Direct Terminal Vocalization:
    .\.venv\Scripts\python speak.py "Construct operational. Standing by for directives, Michael."
  2. Piped STDIN (Pipe ANY CLI output straight into laptop speakers):
    git status | .\.venv\Scripts\python speak.py
Write-Output "Perimeter scan complete. Defensive posture green." | .\.venv\Scripts\python speak.py
  3. Interactive Voice REPL:
    .\.venv\Scripts\python speak.py --interactive
  4. Python Module Import:
    from speak import speak, speak_async
speak("Audio driver online and hooked directly into sound card.")

📡 RF Spectrum Wi-Fi Reconnaissance & Authenticated Connection (wifi_manager.py)

The construct includes a dedicated physical-layer 802.11 Wi-Fi management subsystem capable of in-flight radio frequency reconnaissance, cryptographic profile synthesis, and association handshakes:

1. In-Flight Spectrum Scanning

  • Queries physical WLAN adapters via native netsh wlan show networks mode=bssid.
  • De-duplicates and parses multi-BSSID mesh beacons, radio spectrum bands (2.4 GHz vs 5 GHz vs 6 GHz), RF channels, encryption algorithms (CCMP/TKIP/GCMP), and signal RSSI percentages.
  • Computes tactical visual indicators: 🟢 Connected, 🔒 Secured, 🔓 Open, plus 4-stage signal strength bars (📶 4/4 Excellent, 3/4 Good, 2/4 Fair, 1/4 Weak).

2. Authenticated Association & Handshake Verification

  • Synthesizes dynamic Microsoft WLAN profile XML schemas (http://www.microsoft.com/networking/WLAN/profile/v1) with tailored WPA2PSK / WPA3 / Open key material.
  • Injects profile into Windows WLAN service via netsh wlan add profile.
  • Initiates association handshake with 20-second state polling loop to verify the 4-way WPA cryptographic exchange.
  • Automatically and securely deletes temporary XML profile artifacts from disk immediately upon registration.

3. CLI Command Suite 

# In-flight spectrum scan (prints formatted ASCII tactical matrix)
.\.venv\Scripts\python wifi_manager.py scan

# Check active link status, BSSID, RSSI %, band & link rates
.\.venv\Scripts\python wifi_manager.py status

# Associate to target network with passphrase
.\.venv\Scripts\python wifi_manager.py connect "Olalo-5G" "2011@Olalo"

# Disconnect WLAN interface
.\.venv\Scripts\python wifi_manager.py disconnect

4. Live Cyber HUD Modal (#wifi-scanner-modal)

  • Radar Laser Sweep Animation: High-tech visual scanner overlay when refreshing spectrum.
  • Active Connection Card: Real-time connected SSID, band, signal badge, and link status.
  • One-Click Connect Drawer: Click any discovered network to pop up the association drawer, enter the password (with instant show/hide toggle), and trigger authenticated connection in real time.
  • Quick Access: Triggered directly via 📡 Scan Nearby Wi-Fi or 🔑 Connect Wi-Fi terminal chips or the Device Inspector action matrix.

📺 Local Wi-Fi & Smart TV Command Matrix

60-Second Setup for Smart TVs & Laptops:

  1. Android TV / Google TV / Sony / TCL / Hisense / Fire TV:
    • Go to TV Settings -> Device Preferences -> About -> click Build 7 times to enable Developer Options.
    • Turn on Network Debugging / Wireless ADB.
    • Note the TV's IP (e.g. 192.168.1.105).
    • Tell Cortana on Telegram or Web HUD:
      "Register my Living Room TV as an Android TV at 192.168.1.105."
  2. Roku TV:
    • Note the Roku TV's IP. Cortana commands it via native ECP HTTP REST on port 8060 (power, mute, volume, home).
  3. Laptops & PCs:
    • Enable Wake-on-LAN (WoL) in BIOS/adapter properties.
    • Tell Cortana:
      "Register my Main Laptop as a LAPTOP-PC at 192.168.1.10 with MAC AA:BB:CC:DD:EE:FF."

⚡ HALO Workspace Bridge & Local CLI Daemon (halo_agent.py)

Option 2 architectural implementation: an autonomous principal systems engineer and secure local CLI/workspace bridge connecting your terminal, desktop browser, and mobile devices directly to your files:

1. Three Ways to Execute Refactors & Directives

A. Instant Terminal CLI (Interactive Unified Diff + Confirmation)

Run directly inside your project directory to inspect colorized diffs and confirm [y/N] before modifying any file on disk:

.\.venv\Scripts\python halo_agent.py "patch src/components/Dashboard.tsx : memoize heavy state selectors and remove redundant re-renders"

B. Autonomous Zero-Prompt Mode (HALO_AUTO_APPLY=1)

Start the background daemon with auto-apply enabled so remote directives from your phone or Web HUD automatically create a non-destructive git stash + snapshot backup, apply the patch, verify syntax, and report the diff back:

$env:HALO_AUTO_APPLY="1"
.\.venv\Scripts\python halo_agent.py

C. Dual-Route Web HUD Dispatcher (dispatchDirective)

Your live Web HUD automatically tries the local daemon bridge (http://127.0.0.1:5050/execute) first for zero-latency execution on desktop, and automatically routes through the Supabase command_queue table when you are on your mobile phone without port forwarding.

2. Core Architectural Mechanics

  • Path-Sandboxed Security: Restricts all read/write operations to WORKSPACE_ROOT via Path.resolve().is_relative_to() to block directory traversal (../) and unauthorized system file access.
  • Git Stash + Snapshot Backup Engine: Creates timestamped backups in .halo_backups/ and non-destructive git stash create entries prior to touching files.
  • Syntax Verification Rollback: Runs automated syntax checks (py_compile for Python, node --check for JavaScript) post-write and restores the backup snapshot automatically if compilation fails.
  • Resilient Model Cascade: Seamlessly falls back from Google Gemini Flash/Pro to Groq Qwen/Llama on rate limits or API spikes.

🌐 Live Web Command HUD (Vercel)

Production URL: https://telemetry-news-weather.vercel.app

To deploy updates directly from PowerShell:

npx vercel --prod --yes

On your live Vercel site:

  1. Click Settings (⚙️) in the top navigation bar.
  2. Enter your Supabase URL and Anon Public Key.
  3. Click Save & Connect HUD.
  4. Click VOICE COMM (MIC) to speak directly to Cortana through your browser with real-time speech recognition!
  5. Click VOCALIZE SITREP to hear Cortana read the complete strategic briefing aloud.
  6. Monitor the Local Wi-Fi & LAN Device Mesh grid for live device statuses, IP/MAC addresses, and one-click hardware controls.
Downloads last month

-

Downloads are not tracked for this model. How to track
Inference Providers NEW
This model isn't deployed by any Inference Provider. 🙋 Ask for provider support