BugFlow Tracker

Live Demo: https://bugflow-tracker.onrender.com

BugFlow is a lean, AI-powered bug tracking system designed for speed and clarity. It helps teams report issues and leverages LLMs to automatically triage them by category and severity.

Features

• Modern UI: A clean, Linear-inspired dashboard built with React and Tailwind CSS.
• AI Triage: Automatically suggests bug category (UI, Backend, Performance) and severity (Low, Medium, High) using Google Gemini.
• State Enforcement: Strict workflow transitions (OPEN → IN_PROGRESS → RESOLVED → CLOSED).
• Mandatory Documentation: Requires resolution notes before a bug can be closed.

Tech Stack

• Frontend: React 19 (Vite), Tailwind CSS 4
• Backend: Python (Flask)
• Database: SQLite (SQLAlchemy)
• AI: Google Gemini Pro API
• Production: Gunicorn (Web Server)

Key Decisions

• Unified Service: To simplify deployment and reduce complexity, the Flask backend serves the compiled React production build.
• Absolute Paths: Used for SQLite database location to ensure data persistence across different deployment environments.
• Smart Fallback: Implemented a keyword-based triage layer that takes over if the AI model is unavailable or returns invalid data.
• Pragmatic Architecture: Avoided over-engineering by keeping logic concentrated in a few well-defined files (app.py, services.py, models.py).

Safe AI Usage

• Temperature Control: Set to 0.2 to ensure deterministic and predictable triaging results.
• Validation Layer: All AI outputs are validated against strictly defined schemas before being saved.
• Privacy: Only the bug title and description are sent to the LLM; no user or system metadata is exposed.

Known Limitations

• File System: Uses a local SQLite file. For high-availability production, an external database (PostgreSQL) would be recommended.
• Authentication: Intentionally omitted to maintain the "lean" philosophy of the initial assessment project.

Local Development

1. Backend:
   • pip install -r backend/requirements.txt
   • flask run (ensure GEMINI_API_KEY is in backend/.env)
2. Frontend:
   • cd frontend && npm install && npm run dev

Evaluation Criteria (How this system meets them)

• Structure: Clear separation between UI (React) and Logic (Flask). Backend logic is logically grouped by concern (models, schemas, services).
• Simplicity: Avoided complex state management or heavy middleware. Standard REST patterns make the code predictable.
• Correctness: The BugService strictly enforces business rules, preventing invalid status transitions and ensuring closure requirements are met.
• Interface Safety: Pydantic models guard every entry point, ensuring types are correct before they touch the database.
• Verification: Automated tests in backend/tests/test_bugs.py prove that core rules (like status flow) remain intact.
• Observability: Added debug logging for AI triaging and comprehensive error handling for malformed JSON responses.
• AI Usage: Gemini Pro is used strategically for data enrichment (triage), with a robust validation layer and a deterministic fallback heuristic.