Projects / UofTHacks 13

Award winner1Password: Best Security HackConfidence: low

Alias

Upload an executable and get readable code, its behaviours and a safety verdict before you run it.

Original submission ↗View source ↗Open demo ↗Watch video ↗Use as a reference →
Project facts
Event
UofTHacks 13
17–18 Jan 2026
Team
3 people
Build
software only
AI
llm api, ml model

Problem

Antivirus warnings say a file may be unsafe but not what it does, so users cannot judge them.

Solution

Decompiles the binary, converts pseudo-C to readable C with a specialised model, cleans it with an LLM, extracts strings and behaviours such as network or file access, and explains them with a safe, suspicious or unsafe verdict in a browser extension.

Demo moment

Team-described: drop in an .exe and read what it would do.

Why it stood out

  • organizer statedWon “1Password: Best Security Hack” among 160 submissions.Source ↗
  • team statedExplains behaviour instead of only labelling files.Source ↗
  • curator inferenceDecompilation plus explanation is real security depth for a student build.

Lessons for your next hackathon

  • In security, an explanation users can check beats a bare verdict.

Similar situations

Projects that share domains, mechanisms or judging lenses — often with a different problem and stack.Tune the comparison.

Award winner · 2025

YeetThePacket

Turns packet captures into plain-English, evidence-backed security narratives.

Shared: domains security networking · mechanisms cited evidence trust · lenses technical depth · ai llm api · stack python

Award winner · 2026

BiasLens

Upload a trade log to see scores for overtrading, loss aversion and revenge trading, with explanations and coaching.

Shared: mechanisms cited evidence trust, visible engineering depth · lenses technical depth · ai ml model, llm api · stack python, next.js, gemini

Award winner · 2025

Superhuman

A swarm of AI agents searches CCTV footage in parallel, restores damaged frames and matches suspects from text descriptions.

Shared: domains security networking · mechanisms visible engineering depth · lenses technical depth · stack python, next.js

Award winner · 2025

Forexery

A reinforcement-learning agent that trades USD/JPY using news sentiment plus technical signals.

Shared: mechanisms visible engineering depth · lenses technical depth · ai ml model, llm api · stack python

Sources

  1. https://devpost.com/software/decompiler · devpost project · retrieved 2026-10-13
  2. https://uofthacks-13.devpost.com/project-gallery · devpost gallery · retrieved 2026-10-13