Capabilities

Everything inside the command layer

Nine modular subsystems engineered on a zero-trust, local-first transport - from WebRTC optical surveillance and AES-GCM-256 PIN pairing to acoustic voice intercom, MCP hardware tools, and 60Hz delta input.

Tactical Eye WebRTC surveillance matrix featuring PC display and optical webcam sensor feeds — Tactical Eye Vision - low-latency WebRTC dual-stream surveillance matrix

01 · Real-time WebRTC

Tactical Eye & Surveillance Matrix

Engineered on Google WebRTC SDK (Android) and Python aiortc backend, Tactical Eye transforms your handset into a live optical surveillance cockpit. Stream your PC desktop display and optical webcam sensors in real time under DTLS-SRTP encryption.

Features non-blocking camera hot-swapping and automated ICE candidate rewriting for Android tethered hotspot networks, with instantaneous toggle between video surveillance and low-latency acoustic voice transmission.

WebRTC DTLS-SRTP encrypted stream · 0% idle CPU overhead
Dual live feeds: PC desktop stream + optical webcam sensor
Low-latency acoustic voice mode with live transmission ring
Automated ICE candidate rewriting for Android hotspot networks

Synchronized 6-digit PIN pairing and QR gateway screen with zero-trust AES-GCM-256 encryption — Zero-Trust Pairing - 6-digit PIN gateway and AES-GCM-256 cryptographic link

02 · Zero-Trust Security

AES-GCM-256 Cryptographic Core & PIN Pairing

Every socket frame and REST payload is encrypted end-to-end using AES-GCM-256 with 12-byte IVs, 128-bit authentication tags, and PBKDF2 HMAC-SHA256 derived keys, eradicating legacy plaintext passkeys.

Pair new handsets in seconds using a synchronized 6-digit PIN gateway with auto-focus digit slots and 5-minute expiry windows. Manage trusted endpoints via a JSON registry with granular permission tiers (including Shell God-Mode opt-in).

AES-GCM-256 authenticated encryption across Android and Python
Synchronized 6-digit auto-focus PIN pairing gateway
Persistent trusted device registry with one-click revocation
Token-bucket rate limiter: 3 failed attempts = 10-minute ban

Audio subsystem screen with push to talk control — Voice subsystem - push-to-talk acoustic intercom

03 · Acoustic Intercom

Mobile-to-PC Live Voice Intercom & PTT Engine

Turn your Android device into a live hardware walkie-talkie for your machine. Audio is captured at native 44,100 Hz / 16-bit PCM, converted with zero-padding Rule-of-6 Base64 encoding, and streamed over the persistent socket directly to the PC's pyaudio thread.

Operates in a dedicated, isolated deck with glitch-free graphicsLayer mic pulses and haptic milestones, giving you instantaneous voice communication into your workspace with zero background CPU strain.

16-bit PCM mono hardware capture at 44.1 kHz
Rule-of-6 zero-padding Base64 audio streaming pipeline
Direct PC speaker playback via thread-safe pyaudio queue
Zero PC load · instant push-to-talk physics

Expansion modules sidebar listing available VISAR modules and MCP gateway status — Expansion bay - modular decks, MCP gateway and device controls

04 · Model Context Protocol

MCP Hardware Gateway for AI Agents

VISAR Edge Mobile runs a full Model Context Protocol server backed by a pure-OS Android tool engine. Cursor, Claude Desktop, Claude Code, Roo Code or a custom Python agent can list installed apps, launch them, read the latest OTP, set the clipboard, ping the device, share payloads and pull live power diagnostics.

Requests travel through an asynchronous Hold & Wake bridge: the server opens a UUID-keyed future, broadcasts the call over WebSocket, and resolves it when the handset replies - with a five-second timeout, guaranteed cleanup and per-tool feature flags.

8 hardware tools · 3 live resources for Cursor & Claude
Zero-cloud local WebSocket bridge with Hold & Wake futures
Structured error codes: DEVICE_OFFLINE, PERMISSION_DENIED, TOOL_DISABLED
Non-disableable execution banner on every AI-triggered action

VISAR Edge Mobile AI core chat screen with real-time reasoning stream — Intelligence hub - Groq multi-model reasoning and live thinking streams

05 · Neural Intelligence

Groq Multi-Model Engine & Live Thinking Stream

The neural intelligence hub lets you switch between state-of-the-art open-weights models and your local PC agent on the fly, streaming internal reasoning tokens in real time inside a pulsing cyberpunk thinking accordion.

Maintains conversational context across sliding windows, renders multiplatform markdown code blocks, and accepts push-to-talk voice capture directly from the handset.

Dynamic model switcher: Llama 3.3 70B, DeepSeek R1, Qwen 2.5 32B
Real-time SSE <think> reasoning token streaming accordion
Multiplatform markdown rendering with code block highlighting
Context-aware chat payloads over persistent local socket

Dual-node telemetry screen comparing PC host and mobile edge metrics — Dual-node telemetry - PC host and mobile edge side by side

06 · Observability

Dual-Node Live Telemetry & Ampere Engine

The telemetry hub streams CPU load, memory, disk, power state and top processes from the host, side-by-side with battery thermals, memory, voltage and storage readings from the handset.

A native ampere engine reads live current, drain rate, remaining capacity and time-to-full with multi-fallback kernel readers that eliminate the 0mA/0.004V OEM anomalies, backed by a Tier-2 Ephemeral Gating engine for 0.0% idle CPU.

Comparative PC host and mobile edge live telemetry matrix
Native Ampere engine: live mA current, drain rate and capacity
Tier-2 Ephemeral Gated CPU engine: 0.0% idle PC overhead
Top-10 process monitor with PID search and two-step process kill

VISAR Edge Mobile tactical grid screen with app launchers and system controls — Tactical grid - one tap runs anything on the host machine

07 · Remote Execution

Tactical Command Deck & Vector Macro Grid

Build your own launcher grid with Material 3 vector icons, then trigger builds, scripts, IDEs or media from your phone. Long-press any tile to customize or override icons, backed by persistent Jetpack DataStore storage.

The Ghost Typist panel injects batch text with a 5-item persistent history vault, accompanied by system volume, screen lock, and a deliberate slide-to-confirm shutdown control for the entire ecosystem.

Customizable tactical macro grid with Material 3 vector icons
Ghost Typist text injection with 5-item sliding command history
Raw terminal runner with live server acknowledgements
Deliberate slide-to-confirm on destructive power actions

Orbital drop pod file uplink screen — Drop pod - chunked zero-RAM multi-file transfer to the host

08 · Transfer Layer

Zero-RAM Drop Pod & 2-Way Clipboard Sync

Share anything from Android directly to the host machine - the app registers as a headless OS share target, slicing massive files (e.g. 4K video) into chunked streams with atomic `.part` reassembly and instant resume capability.

Copy an image on the PC and it arrives on the handset through a zero-overhead 16x16 hash-deduplicated sync pipeline; share a web URL from your phone and it opens instantly in your PC's default browser.

Chunked zero-RAM streaming engine with atomic `.part` assembly
Headless Android share sheet target (single & multi-file)
PC-to-mobile image sync with 0% idle CPU grab pipeline
Smart URL extraction: shared links auto-open in PC browser

Holographic trackpad screen with click and keyboard controls — Holographic trackpad - 60Hz delta input with gestures

09 · Input & Navigation

Holographic Trackpad & Universal Adaptive Dock

Pointer movement is streamed as filtered 60Hz deltas with a tuned deadzone, two-finger natural scrolling, and double-tap-to-drag gestures behaving like a physical laptop trackpad.

The universal adaptive navigation engine allows you to dynamically pin, unpin, reorder, and edit 4 to 7 navigation tabs directly from settings, complete with tactile haptic milestones.

60Hz delta input pipeline with micro-scroll deadzone filter
Two-finger natural scrolling and native double-tap-to-drag
Universal adaptive dock builder: customize 4 to 7 tabs dynamically
Zero-allocation canvas rendering using drawWithCache

Ecosystem Depth

And Many More Engineered Capabilities

Purpose-built subsystems delivering zero-compromise performance, desktop automation, and security.

FastAPI Cyber Web Dashboard

A full desktop browser control center at http://localhost:8000/dashboard with real-time gauges, terminal runner, and Drop-Pod staging manager.

3-Tier Cinematic Boot Engine

Tactical, Combat, and Cinematic boot modes with rotating hexagonal HUD reticles, 30ms decryption scrambler, and synchronized haptics.

Zero-Residue Staging Sterilization

Startup cleanup hook in the FastAPI lifespan automatically wipes residual temp files on boot, guaranteeing zero disk bloat.

Intelligent URL Browser Routing

Shared links from Android (YouTube, GitHub, Twitter) are extracted and automatically launched in your PC's default browser.

Anti-Ghost Native System Tray

Embedded pystray daemon running concurrently with the server ensures the PC companion is always accessible and easily managed.

Tier-2 Ephemeral Gated CPU

PC background daemon operates at 0.0% CPU overhead while idle, activating heavy process scanning strictly when breakdown views are requested.

Under the hood

Engine Architecture & Stack

Language: Kotlin 1.9+ (Jetpack Compose) · Python 3.10+
Real-Time Video/Audio: Google WebRTC SDK 125 · aiortc · DTLS-SRTP · pyaudio
Cryptography & Security: AES-GCM-256 · PBKDF2 HMAC-SHA256 · PIN Gateway
AI & Protocols: MCP SDK 1.28+ · Groq API (Llama 3.3/DeepSeek R1/Qwen) · FastAPI
Architecture: MVVM · 4-Tier Settings · Hold & Wake Bridge · Ephemeral Gates
Networking: WebSockets · UDP Zero-Conf · OkHttp · Cyber Web Dashboard

Want it on your device?

Join the early access waitlist to receive the APK build and setup guide the moment it ships.