uplan 1.0.2

📋uPlan - AI-powered development planning automation tool that generates structured project documents and to-do lists. Streamlines your planning workflow with template-based questions, efficient AI usage, and TOML-formatted outputs. Supports multiple AI models and offers customizable templates for development planning.

📋uPlan

uPlan is a Python package that utilizes AI to generate structured development plans and to-do lists.

English | 한국어

📖 Introduction

uPlan automates the development planning process to generate consistent and structured project documentation.

Problems with existing AI chat-based planning:

• Questions and methods vary with each session
• Performance degradation due to increased context as conversations accumulate
• Lack of repeatable workflow

uPlan's Solution:

• Template-based structured questions
• Efficient resource usage by calling AI only when necessary
• Compatibility ensured through structured TOML format output

✨ Key Features

• 🎯 Automatic development plan generation using AI
• ✅ Detailed to-do list creation based on plans
• 📝 Structured output in TOML format
• 🔄 Interactive template customization
• 🛠️ Support for various AI models (OpenAI, Anthropic, Gemini, Deepseek, Ollama, etc.)

🔄 How It Works

uPlan operates through the following workflow:

1. Structured Question Generation (Code): Creates questions based on user-provided templates
2. Question Response (User): Provides answers to structured questions
3. Plan Generation (AI): Creates development plans based on user responses
4. Plan Verification (User): Reviews and approves the generated plan
5. To-Do List Generation (AI): Creates detailed to-do lists based on the approved plan
6. Final Verification (User): Final review and approval of the to-do list

This process yields optimal development plans through efficient interaction between code (automation), user (decision-making), and AI (generation).

🚀 Quick Start

Installation

pip install uplan

Run with Default Settings

uplan

The default model is ollama/qwq.

Specify a Particular Model

uplan --model gemini/gemini-2.0-flash-thinking-exp

Add the GEMINI_API_KEY key-value pair to your .env file. You can get a free key here.

🤖 Supported Models

For more details, refer to MODELS.md.

📋 Detailed Usage

uPlan supports the following command structure:

uplan [global options] [command] [command options]

Global Options

Option	Description	Default
--model	LLM model to use	"ollama/qwq"
--retry	Maximum retry count for LLM requests	5
--category	Template type	"dev"
--input	Input template folder path	"./input"
--output	Output file save folder	"./output"
--debug	Enable debug mode	false

Output Files

The following files are generated as a result of execution:

• plan.toml: Development plan document
• todo.toml: To-do list
• todo.md: To-do list in markdown format
• todo.json: To-do list in checklist format (including completion status)

Commands

Basic Execution (Plan Creation)

Running without a command operates in plan generation mode:

uplan [global options]

Examples:

# Run with default model and dev category
uplan

# Specify model and category
uplan --model "ollama/qwq" --category "custom"

# Change input/output paths
uplan --input "./my-templates" --output "./my-plans"

Note: If templates don't exist in the specified --input/[category] path, they will be automatically initialized.

init - Template Initialization

Creates template files:

uplan init [template] [--force]

Options:

• template: Template name to initialize (default: "dev")
• --force: Force overwrite of existing files

Examples:

# Initialize default dev template
uplan init

# Initialize custom template
uplan init dev_en

# Force overwrite existing template
uplan init dev --force

🛠️ Template Customization

plan.toml

A template that includes prompts and Q&A structure for basic planning.

[prompt]
role = "You are a good code architect and have a good understanding of the development process."
goal = "Create a plan for development."
preferred_language = "English"
instructions = [
    "Review what's already entered in <template>.",
    "<select> can contain multiple contents.",
    "Fill in the <select> parts to create the final deliverable."
]
output_structure = [
    "Write it in JSON format inside a ```json ``` codeblock.",
    "Key values use lowercase"
]

Template Question Structure:

[template.project_basics.overview]
ask = "Please describe the overview of the project"
description = "What you are making (app, service, etc.), target platform (web, mobile, desktop, etc.), main users, etc."
required = true

Property	Description
ask	Basic question
description	Additional explanation (AI auto-generates if answer not provided)
required	Whether to present the question (default: false)

todo.toml

A template for generating detailed to-do lists based on the plan.

[template.frontend]
framework = ["<select> (e.g., react, vue, angular)"]
tasks = [
    "<select> (e.g., design login page UI, design sign up page UI, implement user input validation logic)",
]

Property	Description
frameworks	AI specifies based on the content of output/dev/plan.toml
tasks	AI generates specific to-do list based on the content of output/dev/plan.toml

👨‍💻 Contributing

Issues and pull requests are welcome!

📄 License

See the LICENSE file for more details.

---

Made with ❤️ by EasyDev