Code Buddy - project-aware code generation with ChatGPT, Gemini, Ollama, etc.

Code Buddy integrates LLM-based code generation directly into the Unity Editor.

Its primary goal is to accelerate prototyping and development by leveraging your project's existing context. Instead of writing boilerplate or repeatedly explaining your architecture, use Code Buddy to:

Generate context-aware C# scripts, shaders, or UI documents that integrate with your existing code. You can even provide URLs to documentation or examples within your requests.
Modify, refactor, or get explanations for existing code segments.
Discuss code structure, brainstorm approaches, or plan architecture via chat.
Work entirely offline using local models via Ollama.

🚀 Project-Aware Code Generation

Leveraging AI models and your project's context, Code Buddy generates functional C# code and shaders consistent with Unity's code style. It analyzes your existing codebase to ensure generated scripts integrate correctly, reducing the need for manual adjustments and context-setting compared to standard AI assistants.

💬 Integrated Code Chat & Analysis

Use the integrated chat interface to discuss specific code blocks, request refactoring suggestions, or get explanations of complex logic. Useful for understanding inherited code or exploring different implementation approaches.

🏷️ Flexible LLM Usage (No Subscription)

Code Buddy requires a one-time purchase. Use your own API keys (pay-as-you-go model) for services like OpenAI, DeepSeek, Gemini, Claude AI, or install Ollama to run supported LLMs locally for free, enabling fully offline use. You control the provider and costs.

🎮 Unity-first workflow

Perform code generation, modification, and analysis tasks directly within the Unity Editor UI, minimizing workflow disruption. Generated code adheres to standard Unity C# conventions for readability and maintainability.

🎓 Accessible Prompting

Designed to be effective without requiring highly technical or complex prompts for standard tasks. You can iterate on requests and brainstorm via chat before generating the final code, making it suitable for exploring solutions.

🌍 Proxy support

If you live in a region where the LLM provider is not accessible, do not worry. You can redirect Code Buddy to your proxy server or one of the publicly available proxies.

🛡 Local Context Processing & Privacy

All project context analysis occurs locally; your codebase is not uploaded. Only the generated prompts and a minimal project description are sent to the selected external AI provider.

API keys are encrypted locally.

(Using Ollama enables fully offline operation)

⚠️ IMPORTANT ⚠️

Requires user setup of one of the following:

1. OpenAI API Key. It does not work with ChatGPT Plus subscription.

2. DeepSeek API Key.

4. Gemini API Key.

5. Claude AI API Key.

3. Ollama installed locally

JOIN DISCORD | DOCUMENTATION | YOUTUBE

Release Notes

2.7.0

Gemini support
Claude support
Autoinclude content from the links in the request
Small UI fixes

2.5.1

Ability to exclude folders from the project context
New OpenAI models added - gpt-4.5-preview and o1-pro
Bug fix: disappearing "Initialize" button

2.5.0

Ollama support
DeepSeek API support
Independent settings for every provider
Folders drag-and-drop added
Fix Asset reference warning when moving the Code Buddy folder

2.0.3

New project context generator - improved work with asmdefs and abstractions
Fix NullReferenceException when entering play mode with Code Buddy window inactive
Fix o1 and o3 models naming and usage

2.0.0