TaskFlow — AI Feature Generation Demo

A modern task management app built with Next.js 14, React, TypeScript, and Prisma. TaskFlow provides a comprehensive solution for managing tasks, and projects with an intuitive drag-and-drop Kanban board interface.

This repository is intentionally structured as a demo and sandbox for Bitovi's AI workflows:

• Understanding a codebase and generating copilot instructions
• Automatically implementing features from a Jira ticket

Reference materials:

• Instruction generation workflow
• Feature generation workflow
• Example Jira ticket used in this repo
• Figma designs

What you’ll do in this demo

• Run the app locally with seeded sample data
• Explore the baseline experience on the main branch
• Review the AI-completed feature for USER-13 on the user-13-search-and-filter branch
• Compare branches and see exactly what the AI changed
• Optionally, follow the Bitovi workflows to reproduce the feature implementation with your own AI agent

Tech overview

• Next.js 14 App Router, React, TypeScript
• Prisma ORM with SQLite for local development
• shadcn/ui + Radix + Tailwind for UI
• Feature-based structure with server actions and typed components

Installation

Taskflow requires Node.JS version 18 or greater

1. Clone the repository

   git clone https://github.com/bitovi/taskflow
   cd taskflow
   npm install

2. Set up the database

   # Create, migrate and populate the database
   npm run db:setup

   This will create sample users and tasks for testing. Default login credentials:

   • Email: [email protected]
   • Password: password123

3. Start the development server

   npm run dev

   Then open http://localhost:3000

Branches used in this demo

• main

  • Baseline application used as the starting point for AI feature work
  • Visit /tasks to see the tasks page before the feature is implemented

• user-13-search-and-filter

  • Contains the implementation of Jira ticket USER-13
  • Adds a searchable task bar and filter controls to the /tasks page
  • Follows the project’s patterns (server actions, Prisma, shadcn/ui, accessibility)

Common Git operations for exploring the demo:

# Fetch all branches
git fetch --all

# Switch between the baseline and AI-implemented feature
git switch main
git switch user-13-search-and-filter

# See what changed between branches
git diff main...user-13-search-and-filter

The AI-implemented feature (USER-13)

Ticket: https://bitovi-training.atlassian.net/browse/USER-13

Feature summary:

• Adds a searchable input and filters on the /tasks page
• Lets users quickly find tasks by text, priority, and status

How to try it:

1. Start the app (see Installation above)
2. Log in with the seeded account or create your own
   • ([email protected] / password123)
3. On main, navigate to /tasks and note the baseline behavior
4. Switch to user-13-search-and-filter and refresh /tasks
5. Try searching by task title/description and filtering by priority/status

Tutorial: Running the AI workflows yourself

If you want to recreate the experience with your own AI agent, follow the Bitovi guides:

1. Instruction generation

• Goal: produce a codebase-specific instruction file the AI will follow when writing code
• Guide: https://github.com/bitovi/ai-enablement-prompts/tree/main/understanding-code/instruction-generation

2. Feature generation

• Goal: point your AI at a Jira ticket (e.g., USER-13) and have it implement the feature
• Guide: https://github.com/bitovi/ai-enablement-prompts/tree/main/writing-code/generate-feature

Suggested flow:

• Start on main
• Provide your agent with the instruction generation prompt to build a coding conventions file
• Provide your agent with the feature generation prompt and the USER-13 ticket
• Let the agent create a feature branch and implement the change
• Compare your branch to user-13-search-and-filter to see how close you match

Database management

Available scripts:

• npm run db:seed — Populate the database with sample data
• npm run db:clear — Clear all data from the database
• npm run db:reset — Clear and re-seed the database

The seed script creates:

• 7 sample users with different roles and profiles
• 30+ sample tasks with various priorities, statuses, and assignments
• Realistic task data including descriptions, due dates, and assignments

Local development database:

• Uses SQLite via Prisma (see prisma/schema.prisma)
• Seeded data is safe to reset at any time using the scripts above

Contributing

1. Fork the repository
2. Create a feature branch
3. Follow the established code patterns
4. Test your changes thoroughly
5. Submit a pull request

License

This project is licensed under the MIT License.