v1.0.0
EN DE
GitHub ← About Us
Documentation · Current

KI DEVment Documentation

Complete guide to installation, configuration and usage of the KI DEVment platform – from first steps to deployment.

📖 Welcome to KI DEVment

KI DEVment is a Python-based platform that simulates a fully automated AI development team. It coordinates 10 specialized AI agents that work together to automatically plan, develop, test and deploy your software projects.

This documentation guides you through every aspect of the platform – from installation to configuration to production use.

💡

Tip: If you just want to get started quickly, jump directly to the section Quick Start.

📥 Installation

System Requirements

  • Operating System: Windows 10/11, macOS or Linux
  • Python: Version 3.10 or higher (3.11+ recommended)
  • RAM: At least 4 GB (8 GB recommended)
  • Storage: 500 MB free (1 GB recommended)
  • Ollama (optional): For local AI – at least 2 GB additional RAM

Step 1 – Download & Directory

Terminal
# Change directory
cd "KI DEVment"

Step 2 – Install dependencies

Terminal
pip install -r requirements.txt

Step 3 – Install Ollama (optional)

Ollama allows running AI models completely locally – no internet, no costs, maximum privacy.

ℹ️

Visit ollama.ai.

Terminal – Download model
ollama pull llama3
ollama pull mistral
ollama pull codellama

Step 4 – Start application

Terminal
python main.py

Compile as EXE (Windows)

Terminal
python compile.py
💡

Tip: The compiled EXE file can be run on the target machine without Python installation.

Quick Start

The fastest way to create your first project with KI DEVment:

  1. Start KI DEVment with python main.py
  2. Describe your project in natural language in the input field
  3. Click "Start Project" – the platform analyzes your description
  4. Watch the agents working in the live dashboard
  5. Project finished! The application was automatically created and can optionally be deployed via FTP
⚠️

Note: Make sure Ollama is running (ollama serve) before starting a project – otherwise the local AI cannot be used.

Example Prompt

Project Description
"Create a simple to-do list web application with MySQL database, a modern frontend and a REST API in the backend."

🏗️ Architecture – Overview

KI DEVment is based on a multi-agent architecture. A central control platform coordinates specialized agents, each with clear responsibilities.

Communication Flow

  1. Input The user describes their project in natural language
  2. Analysis Smart Prompt Analysis evaluates complexity and determines required agents
  3. Orchestration The project coordinator assigns tasks to agents
  4. Development Agents work in parallel and communicate with each other
  5. Testing The beta test agent validates results in the embedded browser
  6. Deployment The FTP manager transfers the finished files to the server
ℹ️

Intelligent Agent Selection Not all 10 agents are activated for every project. The platform only selects the agents actually needed – maximum efficiency, minimum resources.

🤖 AI Agents

The 10 agents form the core of KI DEVment. Each agent has a clear specialization and works independently within its area of responsibility.

AgentFunctionSpecialization
🎯 Project CoordinatorOverall ControlPlanning & Strategy
⚙️ Code GeneratorProgram LogicAlgorithms & Structures
🎨 Frontend DeveloperInterfacesHTML, CSS, JS
🖥️ Backend DeveloperServer LogicAPIs, Data Processing
🗄️ Database ArchitectDB ManagementMySQL, PostgreSQL, SQLite
🧪 Test AutomatorQuality AssuranceUnit & Integration Tests
📊 Production MonitorMonitoringPerformance & Security
📤 FTP ManagerDeploymentUpload & Transfer
🔧 PHP DeveloperPHP GenerationServer-side Scripts
🌐 Beta Test AgentBrowser TestsEnd-to-End Validation

🎯 Project Coordinator

The coordinator is the core of orchestration. It analyzes incoming project requests, creates an action plan, and assigns subtasks to the appropriate agents. It monitors overall progress and escalates if needed.

⚙️ Code Generator

Generates the core logic of an application: algorithms, data structures, and program logic. Works closely with frontend and backend agents to produce consistent code.

🎨 Frontend Developer

Creates user interfaces with HTML5, CSS3, and JavaScript. Supports modern CSS frameworks and responsive design. Passes finished files to the beta test agent.

🖥️ Backend Developer

Develops server-side logic, RESTful APIs, and data processing pipelines. Works with Python and can optionally integrate Node.js components.

🗄️ Database Architect

Designs and manages database structures for MySQL, PostgreSQL, and SQLite. Automatically creates tables, indexes, and relations based on project requirements.

🧪 Test Automator

Creates and executes unit tests and integration tests. Identifies errors and forwards corrections to the responsible agent.

📊 Production Monitor

Monitors the quality and stability of developed software. Checks for performance issues and potential security vulnerabilities before deployment.

📤 FTP Manager

Handles automatic file transfer to the web server. Supports FTP and SFTP with TLS encryption. Can upload individual files or entire directories.

🔧 PHP Developer

Generates PHP files for server-side operations: database connections, table creation, and interfaces. Generated files are automatically transferred via FTP.

🌐 Beta Test Agent

Uses an embedded Chromium browser (via PyQt6 WebEngine) for automated end-to-end testing. Extracts error messages from logs, analyzes problems, and forwards corrections.

🧠 AI Setup

Option A – Ollama (Local, recommended)

Ollama runs AI models completely on your machine – no internet connection, no costs, maximum privacy.

Terminal
ollama serve    # Start server
ollama pull llama3     # Download model
ollama list        # Show available models

Supported Models

  • General purpose language model – General purpose language model
  • Fast & efficient – Fast & efficient
  • Specialized on code – Specialized on code

Option B – OpenAI (Cloud, optional)

If you prefer a cloud-based AI, OpenAI can be configured. An API key is required.

⚠️

Note: When using OpenAI, your data is sent to OpenAI servers. For privacy-sensitive projects, we recommend Ollama.

📤 FTP Configuration

The FTP manager enables automatic file transfer to your web server.

Required Information

  • Server Address: ftp.example.com
  • Username: Your FTP username
  • Password: Your FTP password
  • Port: 21 (FTP) / 22 (SFTP)
  • Target Directory: /var/www/html/
💡

Tip: The transfer uses TLS encryption to protect your credentials. Prefer SFTP if offered by your hosting provider.

🗄️ Database Setup

KI DEVment supports three database systems and can automatically create structures and tables.

Supported Systems

  • MySQL / MariaDB – Ideal for web hosting
  • PostgreSQL – Advanced features & scaling
  • SQLite – File-based, no server needed

Connection Parameters

  • Host: localhost
  • Port: 3306 (MySQL) / 5432 (PostgreSQL)
  • Database: Name of your database
  • User & Password
ℹ️

The PHP developer agent automatically generates PHP files for database connection and table creation – these are transferred directly to your server via FTP.

✍️ Prompt Guide

The quality of your project description directly influences the result. Here are tips for better prompts:

✓ Good Practices

  • Be specific: Describe features, not just the goal
  • Technologie angeben: „mit MySQL" oder „React Frontend"
  • Limit scope: Start with small projects
  • Give examples: "Similar to a to-do list, but with…"

✕ To Avoid

  • Too vague: "Create an app" – too little information
  • Too complex at once: Split large projects into sub-projects
Example – Good Prompt
"Erstelle eine Aufgabenverwaltung mit:
- MySQL-Datenbank für Aufgaben & Benutzer
- REST-API im Backend (Python/Flask)
- Modernes Frontend mit Karten-Layout
- Login/Logout Funktionalität
- FTP-Deployment auf meinen Server"

🔄 Workflow

The typical workflow from input to result:

  1. Prompt Input – You describe your project
  2. Smart Analysis – Complexity is evaluated, agents are selected
  3. Planning Phase – The coordinator creates an action plan
  4. Development Phase – Agents work in parallel on their subtasks
  5. Testing Phase – Automated tests in the embedded browser
  6. Review Phase – The monitor checks quality and security
  7. Deployment (optional) – FTP upload to your server

🚀 Deployment

After development, your project can be automatically deployed to a web server.

Prerequisites

  • An FTP/SFTP access to your web server and FTP configuration entered in KI DEVment

What gets uploaded?

  • HTML/CSS/JS files (frontend), PHP files (database connection, server logic), configuration files, Optional media & assets
💡

Tip: The FTP manager can upload individual files or entire directories at once (bulk upload).

💻 CLI Reference

CommandDescription
python main.pyStart main application
python compile.pyCompile as EXE
pip install -r requirements.txtInstall dependencies
ollama serveStart Ollama server
ollama pull [model]Download AI model
ollama listShow available models

🔧 Troubleshooting

Ollama not found

⚠️

Solution: Make sure Ollama is installed and the server is running (ollama serve). Check that the service is active on port 11434.

FTP connection failed

⚠️

Solution: Check server address, username, password, and port. Make sure your hosting provider has enabled FTP/SFTP.

Python error on startup

⚠️

Solution: Run pip install -r requirements.txt again. Make sure Python 3.10+ is installed.

Beta test agent not working

⚠️

Solution: The beta test agent requires PyQt6 with WebEngine. Check if PyQt6-WebEngine is included in requirements.txt and was installed correctly.

FAQ

Can I use KI DEVment without Ollama?

Yes, you can configure OpenAI as an alternative AI source. An API key and internet connection are required.

How long does a typical project take?

It varies depending on complexity. Simple websites can be ready in a few minutes, more complex applications take longer – but always significantly faster than manual development.

Can I customize the generated files?

Absolutely! All generated files are stored locally and can be freely edited. You can continue developing the project at any time.

Which languages does the interface support?

The user interface is available in German and English. The language can be changed in the settings.