BugTrace AI

• 

  scan page

• 

  investigation page

• 

  fix plan page

• 

  export page

• 

  home page

• 

  analysis page

InspirationBugTrace AI

The Problem :

Every developer has faced this: “It was working yesterday… now it’s broken.” Finding the "first bad commit" in a regression is slow, manual, and frustrating. Developers often: Scan dozens of commits, Rerun tests repeatedly, Guess where things went wrong

Under time pressure, this becomes even harder.

My Solution :

"BugTrace AI is a local-first debugging assistant that automatically finds the first breaking commit and explains why it failed" Instead of manually searching, it:

• Scans commit history
• Detects the passing → failing transition
• Uses an optimized bisect-style approach
• Pinpoints the exact commit responsible

What Makes It Special : BugTrace AI is not just a debugging tool — it is a "debugging + explanation system"

Engineering Accuracy = Works on real repositories, Uses actual Git history, Produces reproducible results

How We Built It :

We designed a modular Python architecture where the core engine powers both CLI and UI.

Key components:

Regression Scanner → finds failure window Investigation Engine → isolates bad commit Commit Analyzer → inspects diffs & changes Fix Plan Generator → suggests next steps HTML Report Generator → shareable insights Streamlit UI → smooth demo experience Optional AI Layer → enhances explanations

I combine:

• Deterministic Git-based debugging
• AI-assisted explanations

Result: Trust + Clarity

Challenges :

• Handling real repository inconsistencies
• Making regression checks reproducible
• Dealing with flaky outputs
• Simplifying complex diffs
• Balancing technical depth with clean UI

What i Learned

• Debugging tools must be clear, not just powerful
• Separation of logic and UI is critical
• AI should support evidence, not replace it

Impact

Developers → debug faster Teams → understand failures clearly Judges → see value instantly

Built With

• api
• git
• llm
• python
• streamlite