All projects

Cybersecurity · Bitcoin

BitShield

A team-built Bitcoin security dashboard; I built the guardian engine and Cogcoin anchoring and co-built its rule-based quantum-exposure scanner.

Context
MIT Bitcoin Expo 2026, Team ACE · hackathon project
Stack
TypeScript, bitcoinjs-lib, Hono, Bun

How it works

Recorded demo

Recorded run of BitShield's rule-based quantum-exposure scanner on public test addresses (no wallet, no API keys). The 3D view renders slowly on a software renderer, so typing looks slow.

MIT Bitcoin Expo 2026 Hackathon, Team ACE. The scanner was co-built with a teammate; my part was the Guardian and Cogcoin work.

  1. Explore the risk view

    The scanner page opens on a 3D quantum-risk visualizer that can be dragged to rotate.

  2. Open a risk category

    Clicking the Spent P2WPKH chip opens its detail panel.

  3. Scan a public test address

    A BIP-173 test-vector address is scanned; the rule-based scanner returns LOW risk for a P2WPKH address.

  4. Scan a Taproot test address

    A BIP-350 test-vector address is scanned and returns MEDIUM risk for a P2TR address.

  5. Read the network panel

    The network-vulnerability panel and the attack countdown sit below the scan results.

Highlights

  • Co-built scanner: classifies six Bitcoin address types (P2PK through P2WSH) by quantum exposure
  • Address typing via bitcoinjs-lib and spent-status lookups via mempool.space Signet
  • Generates a migration plan for at-risk addresses

How it was built

At the MIT Bitcoin Expo hackathon our team built a Bitcoin security dashboard. I wrote the guardian engine, which sends a user's request to Claude and accepts only strict JSON naming one of eight wallet intents, with whitelist validation and a safe "unknown" default. I also wrote the Cogcoin integration that anchors guardian actions on chain. With a teammate, I co-built the quantum-exposure scanner, which types Bitcoin addresses with bitcoinjs-lib and looks up spent status on the Signet network. Teammates built the wallet and Lightning features, dashboard and deployment. Limits: rule-based scanner, not a model; not an autonomous agent.