AutoDebug AI is a fully autonomous, local debugging environment that executes broken code, analyzes runtime signals, and iteratively repairs logic errors without sending a single byte of data to the cloud.
Track: DevForge Problem Statement 4 (Local AI-Supervised Autonomous Debugging Sandbox)
Debugging manually is slow and frustrating. Beginners struggle to interpret stack traces, and fixing one bug often reveals another. AutoDebug AI solves this by implementing a loop that runs entirely on your machine.
🔄 Autonomous Repair Loop: Unlike standard chatbots, this system runs your code in a sandbox. If the fix fails, it captures the new error and retries up to 3 times automatically.
⚖️ Visual Side-by-Side Diffs: Instantly see exactly what changed between your broken code and the fixed version.
🔒 100% Local & Private: Powered by qwen2.5-coder:1.5b via Ollama. No API keys, no cloud latency, full data privacy.
🛡️ Safety First: Automatically creates .bak backups before applying any patches.
🧠 Context-Aware: Understands both runtime errors (stderr) and logic errors (wrong stdout) based on user prompts.
Frontend: Streamlit (Python-based Reactive UI)
AI Engine: Ollama (running qwen2.5-coder:1.5b)
Sandbox: Python subprocess with timeout constraints (preventing infinite loops).
Diffing Engine: Python difflib for generating structured patch instructions.
Prerequisites
Python 3.8+
Ollama (installed and running)
Step 1: Install Dependencies
pip install streamlit requests
Step 2: Setup Local AI
Make sure Ollama is installed, then pull the lightweight coding model:
ollama pull qwen2.5-coder:1.5b
Step 3: Run the Engine
Start the Ollama server (if not running):
ollama serve
Launch the AutoDebug UI:
streamlit run app.py
Select a File: Use the sidebar to choose a Python script from your directory.
Describe the Issue: In the chat box, type the problem (e.g., "This binary search hangs forever" or "Fix the recursion depth error").
Watch the Loop:
The system puts the code in a sandbox.
It detects the error (Crash or Timeout).
It consults the Local AI.
It applies the patch and re-runs the code to verify.
Review: If successful, you see a green "Fix Applied" message and a visual comparison of the changes.
We stress-tested the system against the following common bugs:
🔮 Future Roadmap
Security Static Analysis: Pre-scan code for dangerous imports (os.system, shutil) before execution.
Multi-File Support: Allow debugging across multiple imported modules.
IDE Extension: Porting the core logic to a VS Code Extension (Prototype CLI available).