AI-DLC-UML (AI-Driven Development Life Cycle with UML Modeling)

AI-DLC-UML modifies AI-DLC to enable AI agents to drive the software development workflow with UML modeling. It is intended for those who want to use UML modeling collaboratively in their design practices, even in AI-driven software development.

Key modifications to AI-DLC include:

• Application-design and functional-design artifacts are created as UML models.¹
• A step has been added to maintain consistency between the design artifacts and the codebase.
• Supported AI agents are limited to Claude Code, Codex CLI, and Gemini CLI.²

flowchart TD
    A((Start)) --> B

    subgraph B["Inception phase"]
        direction TB
        B1["Workspace Detection<br/>(Always)"]
        B2["Reverse Engineering<br/>(Conditional)"]
        B3["Requirements Analysis<br/>(Always)"]
        B4["User Stories<br/>(Conditional)"]
        B5["Workflow Planning<br/>(Always)"]
        B6["Application Design *1<br/>(Conditional)"]
        B7["Units Generation<br/>(Conditional)"]

        B1 --> B2 --> B3 --> B4 --> B5 --> B6 --> B7
    end

    B --> C

    subgraph C["Construction phase"]
        direction TB
        C1[/"Loop Start<br/>(for each unit)"\]
        C2["Functional Design *1<br/>(Conditional)"]
        C3["NFR Requirements Assess<br/>(Conditional)"]
        C4["NFR Design<br/>(Conditional)"]
        C5["Infrastructure Design<br/>(Conditional)"]
        C6["Code Generation<br/>(Always)"]
        C7["Design-Code Sync *2<br/>(Conditional)"]
        C8[\"Loop End"/]
        C9["Build and Test<br/>(Always)"]

        C1 --> C2 --> C3 --> C4 --> C5 --> C6 --> C7 --> C8 --> C9
    end

    C --> D

    subgraph D["Operations phase"]
        direction TB
        D1["Operations<br/>(Placeholder)"]
    end

    D --> E((End))

    style B6 fill:#ffcc99,color:#000
    style C2 fill:#ffcc99,color:#000
    style C7 fill:#ffcc99,color:#000

Tailor this workflow as needed to fit your preferred or required workflow. For example, if the target software system will not be mapped to AWS services, some steps could be modified. The Operations phase is included as a placeholder and should be defined according to your workflow.

Demo

Initial prompt: Using AI-DLC-UML, create a Java desktop app for Tetris
AI agent: Claude Code using Sonnet 4.6
Processing time: 139 minutes

The images below show the created Tetris app and some of the UML diagrams created for it. In this demo, we approved the AI agent's output as-is; only the diagram layout was adjusted manually.

Requirements

• Claude Code, Codex CLI, or Gemini CLI

  Info: Based on our personal experience, as of May 2026, we recommend Claude Code because its UML modeling capabilities appear more advanced than those of the others. Next, we recommend Codex CLI.

• Astah Pro v11.0 or later

• Astah Pro MCP v0.2.2 or later

• Node.js v20 or later

Installation

• Install Claude Code, Codex CLI, or Gemini CLI

• Install Astah Pro

• Install the Astah Pro MCP plugin

  Steps to install: Launch Astah Pro -> drag and drop the Astah Pro MCP JAR file onto the Astah Pro window -> restart Astah Pro.

• Install Node.js

Usage

1. It is recommended to disable unused MCP tools to avoid reducing the AI agent's tool-calling accuracy.

2. Open a terminal and go to your project directory

3. Setup AI-DLC-UML in your project directory

   Run this command to place the AI-DLC-UML folders and files in your project directory.

   npx https://github.com/takaakit/ai-dlc-uml.git

   If the setup fails, manually download this GitHub project and place the following folders and files in your project directory.

   Your project directory
    ├ .aidlc-rule-details
    ├ .claude
    ├ .codex
    ├ .gemini
    ├ .mcp.json
    ├ AGENTS.md
    ├ CLAUDE.md
    └ GEMINI.md
   

   If there are conflicts with existing files, back them up if needed, then replace them.

4. Start Astah Pro

   Note: Launch Astah Pro before starting AI agents, and it should remain open throughout the entire workflow.

5. Start AI agents

   Run the claude, codex, or gemini command in your project directory. A confirmation dialog will pop up on initial connection. Check it and click 'Connect'.

   Recommended AI models:

   • Claude Code: Sonnet 4.6 or higher
   • Codex CLI: GPT-5.5 or higher
   • Gemini CLI: Gemini 3.1 Pro or higher

6. Send a prompt starting with "Using AI-DLC-UML, ..."

   Example prompt: Using AI-DLC-UML, create a Java desktop app for Tetris

   Note: If you interrupt a workflow—for example, because the rate limit has been reached—and resume it later, connect the AI agent to the Astah Pro MCP server before resuming.

Additional Info

• AI agents refer to information provided by OMG to understand UML.
• AI agents follow Agile Modeling guidelines when creating UML models.

License

• The original AI-DLC is licensed under the MIT-0 license and is copyrighted by Amazon.com, Inc. or its affiliates.
• All modifications and new content added by this project are released under the CC0 1.0 Universal (Public Domain) license.

Contributing

If you have a proposed improvement, please open an issue. Note that any improvements submitted will be released under the CC0 license.

Links

• AI-DLC
• Claude Code
• Codex CLI
• Gemini CLI
• Astah Pro
• Astah Pro MCP
• Node.js
• OMG
• Agile Modeling