Gemini Coder

The free 2M context AI coding toolkit

Copy folders and files for chatbots or initialize them hands-free!
Use the free Gemini API for code completions, file refactoring and applying AI-suggested changes.

Documentation

Gemini Coder lets you conveniently copy folders and files for chatbots. With the Connector browser extension you can initalize them hands-free!

The extension uses the same context for built-in API features: code completions and file refactoring. Hit apply changes to integrate AI responses with just a single click.

• 100% free & open source: MIT License
• Versitale: Initializes Gemini, AI Studio and many other popular chatbots
• Private: Does not collect any usage data
• Local: Talks with the web browser via WebSockets
• One of a kind: Lets you use any website for context
• Lightweight: Unpacked build is just ~1MB

You own the context

Other AI coding tools try to "guess" what context matters, often getting it wrong. Gemini Coder works differently:

• You select which folders and files provide relevant context
• You control what examples of coding conventions to include
• You know how much tokens are used in web chats and code completion/file refactoring requests

The result? Unmatched in accuracy, speed and cost AI assistance.

Too many tokens fighting for attention may decrease performance due to being too "distracting", diffusing attention too broadly and decreasing a signal to noise ratio in the features. ~Andrej Karpathy

Supported chatbots

Gemini Coder works with many popular chatbots:

• AI Studio
• Gemini
• OpenRouter
• Open WebUI (self-hosted)
• ChatGPT
• Claude
• GitHub Copilot
• Grok
• DeepSeek
• Mistral
• HuggingChat

Quick start for chat

1. Open the new Gemini Coder view from the activity bar (sparkles icon).
2. Select files/folders for the context.
3. Click copy icon from the toolbar.
4. (optional) Install browser integration for hands-free initializations.

Quick start for FIM completions

1. Get your API key from Google AI Studio.
2. Open VS Code and navigate to settings.
3. Search for "Gemini Coder" and paste your API key.
4. Use Command Palette (Ctrl/Cmd + Shift + P) and type "Code Completion".
5. Bind the command to a keyboard shortcut by opening Keyboard Shortcuts (Ctrl/Cmd+K Ctrl/Cmd+S), searching for Gemini Coder: Code Completion, clicking the + icon, and pressing your preferred key combination (e.g. Ctrl/Cmd+I).

Commands

Code Completions

• Gemini Coder: Code Completion - Get fill-in-the-middle completion using default model.
• Gemini Coder: Code Completion with... - Get fill-in-the-middle completion with model selection.
• Gemini Coder: Code Completion to Clipboard - Copy FIM completion content to clipboard.
• Gemini Coder: Change Default FIM Model - Change default AI model for FIM completions.

Refactoring

• Gemini Coder: Refactor this File - Apply changes based on refactoring instruction.
• Gemini Coder: Refactor this File with... - Refactor with model selection.
• Gemini Coder: Refactor to Clipboard - Copy refactoring content to clipboard.
• Gemini Coder: Change Default Refactoring Model - Change default AI model for refactoring.

Applying Changes

• Gemini Coder: Apply Changes - Apply changes suggested by AI using clipboard content.
• Gemini Coder: Apply Changes with... - Apply changes with model selection.
• Gemini Coder: Apply Changes to Clipboard - Copy apply changes content to clipboard.
• Gemini Coder: Change Default Apply Changes Model - Change default AI model for applying changes.

Chat

• Gemini Coder: Web Chat - Enter instructions and open web chat hands-free.
• Gemini Coder: Chat to Clipboard - Enter instructions and copy to clipboard.

Context

• Gemini Coder: Copy Context - Copy selected files as XML context.

Set up custom model providers

The extension supports OpenAI-API compatible model providers for API features.

  "geminiCoder.providers": [
    {
      "name": "DeepSeek",
      "endpointUrl": "https://api.deepseek.com/v1/chat/completions",
      "apiKey": "[API KEY]",
      "model": "deepseek-chat",
      "temperature": 0,
      "instruction": ""
    },
    {
      "name": "Mistral Large Latest",
      "endpointUrl": "https://api.mistral.ai/v1/chat/completions",
      "apiKey": "[API KEY]",
      "model": "mistral-large-latest",
      "temperature": 0,
      "instruction": ""
    },
  ],

Contributing

All contributions are welcome. Feel free to submit pull requests or create issues and discussions.

License

Copyright (c) 2025 Robert Piosik. MIT License.