Code Archeologist

Result Page

Inspiration

Every company has that one scary folder of code no one wants to touch. It’s old, insecure, and undocumented—but it runs critical systems. We built Code Archeologist to face that fear head-on: understand legacy code instead of avoiding it, and rebuild it cleanly for the future.

What it does

Code Archeologist is an autonomous AI system that modernizes legacy software. You paste in old code, and three AI agents take over: --> Archaeologist analyzes the code and finds security risks --> Architect uses Gemini 3 reasoning to design a modern system --> Builder rewrites the app and runs a self-healing loop, fixing its own bugs until it works

How we built it

The UI is built with Next.js. The intelligence comes from Gemini 3: Flash for fast analysis and generation Pro for deep architectural planning Agents are orchestrated through a state machine so each step builds on the last.

Challenges we ran into

Rate limits were tough. We hit a lot of 429 errors and had to build retry logic and a simulation mode just to keep moving. We also had to carefully constrain the AI to stop it from using nonexistent libraries.

Accomplishments that we're proud of

Watching the system fix its own code for the first time—reading errors, applying fixes, and passing tests without human input—felt like working with a real teammate.