🤖 AI-Powered Code Debugging Assistant

A multi-agent AI system that automatically analyzes, debugs, and corrects Python programs using CrewAI, Code Interpreter, Pydantic, and Streamlit.

---

🎥 Demo

Demo Video

---

📖 Overview

Debugging Python programs often requires identifying syntax, runtime, and logical errors before applying safe corrections.

This project implements a CrewAI-powered multi-agent debugging workflow where specialized AI agents collaborate to analyze code, generate corrections, and verify the debugging process.

Unlike traditional prompt-based implementations, this project introduces a Builder Pattern that separates LLM reasoning from structured artifact construction, producing deterministic and strongly typed workflow outputs.

---

✨ Features

• 🤖 Multi-Agent AI Debugging using CrewAI
• 🔍 Automated Syntax, Runtime & Logical Error Detection
• 🛠 Automatic Code Correction
• 🧠 Code Interpreter Integration
• 📄 Builder Pattern for Structured Artifacts
• 📊 Interactive Streamlit Dashboard
• 📈 Observability & Execution Tracing
• 🔄 Side-by-Side Code Comparison
• 📑 Unified Diff Viewer
• 📦 JSON Trace Export

---

🏗 System Architecture

The application consists of:

• Streamlit Frontend
• Debugging Service
• Sequential CrewAI Workflow
• Analyzer Agent
• Corrector Agent
• Manager Agent
• Builder Components
• Structured Pydantic Artifacts
• Observability Layer

---

⚙ Multi-Agent Workflow

                User Code
                     │
                     ▼
        ┌─────────────────────────┐
        │  Code Analyzer Agent    │
        └─────────────────────────┘
                     │
                     ▼
            AnalysisResponse
                     │
                     ▼
            AnalysisBuilder
                     │
                     ▼
             AnalysisReport
                     │
                     ▼
        ┌─────────────────────────┐
        │ Code Corrector Agent    │
        └─────────────────────────┘
                     │
                     ▼
           CorrectionResponse
                     │
                     ▼
          CorrectionBuilder
                     │
                     ▼
       CorrectedCodeArtifact
                     │
                     ▼
        ┌─────────────────────────┐
        │     Manager Agent       │
        └─────────────────────────┘
                     │
                     ▼
          Verification Summary

---

🧩 Builder Pattern

One of the key architectural improvements in this project is the Builder Pattern.

Instead of asking the LLM to generate complex Pydantic objects directly, each agent only produces a lightweight response schema.

Dedicated Builder classes then construct rich domain artifacts.

LLM
      │
      ▼
AnalysisResponse
      │
      ▼
AnalysisBuilder
      │
      ▼
AnalysisReport

LLM
      │
      ▼
CorrectionResponse
      │
      ▼
CorrectionBuilder
      │
      ▼
CorrectedCodeArtifact

Benefits

• Deterministic artifact construction
• Reduced prompt complexity
• Strong type safety
• Better maintainability
• Fewer schema validation failures

---

📦 Domain Artifacts

The debugging workflow exchanges structured artifacts throughout execution.

SourceCodeArtifact

        │

        ▼

AnalysisReport

        │

        ▼

CorrectedCodeArtifact

        │

        ▼

Verification Summary

---

📊 Streamlit Dashboard

Analysis Dashboard

---

Code Comparison

---

Corrected Code

---

Verification Summary

---

Observability Dashboard

---

📈 Observability

The project includes a lightweight observability framework that records:

• Workflow Events
• Agent Execution
• Task Completion
• Execution Metrics
• JSON Trace Export

These metrics are visualized directly inside the Streamlit application.

---

🛠 Technology Stack

Component	Technology
Framework	CrewAI
Language	Python
LLM	Groq Llama 3.3 70B
Validation	Pydantic v2
UI	Streamlit
Execution	CodeInterpreterTool
Observability	Custom Tracer

---

📁 Project Structure

code-debugging-assistant/

│

├── agents/

├── builders/

├── config/

├── observability/

├── prompts/

├── schemas/

├── services/

├── tasks/

├── utils/

│

├── images/

├── streamlit_app.py

├── crew.py

├── main.py

├── requirements.txt

└── README.md

---

🚀 Installation

Clone the repository

git clone https://github.com/ruthuraraj-ml/ai-code-debugging-assistant.git

Move into the project

cd ai-code-debugging-assistant

Create virtual environment

python -m venv .venv

Activate

Windows

.venv\Scripts\activate

Linux / macOS

source .venv/bin/activate

Install dependencies

pip install -r requirements.txt

Configure environment variables

GROQ_API_KEY=your_api_key

Run

streamlit run streamlit_app.py

---

💡 Key Learnings

This project demonstrates:

• Multi-Agent AI Systems
• CrewAI Workflow Design
• Builder Design Pattern
• Pydantic Domain Modeling
• Observability for AI Applications
• Streamlit Application Development
• Software Engineering for LLM Systems

---

🛣 Roadmap

✅ Version 1.0

• Multi-Agent Workflow
• Builder Pattern
• Structured Artifacts
• Streamlit Dashboard
• Execution Tracing
• Code Diff Viewer

🔄 Version 2.0

• Evaluation Benchmark Suite
• Regression Testing
• Prompt Optimization
• Improved Logical Error Detection
• Verification Artifact
• Multi-language Support

---

👨‍💻 Author

Ruthuraraj R

Assistant Professor – Mechanical Engineering

AI • Machine Learning • Generative AI • Agentic AI Systems

---

⭐ If you found this project interesting, consider giving it a Star.