Skip to content
pq-cybarg edited this page May 29, 2026 · 2 revisions

Tetherand

A reverse-tethering, privacy-chain, and threat-detection app for the Solana Seeker, designed for use at security conferences where the local cellular and Wi-Fi environment can't be trusted.

It does four things:

  1. Reverse-tether to a laptop so the phone uses the laptop's internet instead of its own cellular modem — over USB-ADB, USB-AOA, Bluetooth-RFCOMM, or LAN-TCP.
  2. Route traffic through a privacy chain you compose yourself — WireGuard, Mullvad (with post-quantum tunnels and DAITA traffic shaping), NymVPN's Sphinx mixnet, and Tor with its full bridge stack (obfs4, meek, webtunnel, conjure) including the post-quantum NTor handshake when upstream Arti exposes it.
  3. Detect cellular and Wi-Fi threats on-device — IMSI catchers, evil-twin access points, BLE trackers, app permission drift — and surface them in a Threat tab with a one-tap panic button.
  4. Lock the device down for the conference via a single toggle that captures a pre-event attestation snapshot, freezes the trusted-app baseline, runs a honeypot, arms an accelerometer tamper-watcher, and exposes a four-button incident-response runbook.

There is also a contributory AI Guard layer for AI-era threats (deepfake calls, prompt-injection text, synthetic-media provenance, OSINT exposure), which runs entirely on-device with no cloud LLM calls under any circumstances.

Where to start

Project conventions

  • Every defense has a deterministic primary mechanism — a clear rule, threshold, or heuristic — that drives any consequential action. Local AI classifiers are advisory only.
  • All AI runs on-device. No prompt, classification, or telemetry ever reaches a cloud LLM API under any circumstances.
  • No telemetry. Tetherand never phones home. Models update only through whatever privacy chain the user has active.

These three rules are non-negotiable and not configurable through the UI.

Clone this wiki locally