A code editor with AI assistance. It supports chat, apply changes, autocomplete, and inline gray completions etc. These AI features (extensions) are simple Python scripts, so you can easily modify them, create your own, and quickly try out new code tool ideas.

We call these AI features extensions.

For setting up a development environment and detailed information on using extensions and developing/modifying your own, please see our documentation.

Install or update IDE with a single command (clones/update the repo and builds automatically):

curl -sSL https://raw.githubusercontent.com/notbadai/ide/main/install.sh | bash

The packaged application will be available in the ide/dist/desktop/ directory.

To use the AI features, you need to complete the following setup:

1. Setting Up API Keys

Configure API keys for your AI model providers by navigating to Extensions → Management from the left panel. This will open the configuration YAML file located at ~/.notbadaiide/config.yaml.

Add at least one provider with your API key. You can add one or both providers, and remove any that you don't plan to use:

  - provider: "openrouter"
    key: "<YOUR-KEY>"
    default: true
  - provider: "deepinfra"
    key: "<YOUR-KEY>"

Important: Add an OpenRouter API key to enable the code apply feature. Other provider we support do not have the model we use to apply changes.

Lets you talk to your LLM directly from the editor. You can ask for code explanations, code suggestions, and edits.

You can access this from the left-side pane.

• You can select a Chat extension to run from the dropdown menu.
• In the default chat extension, the context sent to the LLM includes only the currently open tabs.
• You can manually add more context, such as additional files or folders using the + context option.
• The files extension automatically determines the relevant context to include, ignoring open tabs and any manually added context.
• In both extensions, the last 1,000 lines of current terminal are also included in the context sent to the LLM.

Lets you review and apply code suggestions from the LLM. You can see a diff, accept the changes, or discard them with a click.

When the chat provides a code suggestion, it appears inside a code block. Each code block includes an Apply button:

• Click Apply to review a diff between your current file and the suggested code.
• Click Accept to apply the changes (the file will be updated) or Reject to discard the suggestion (in the diff view).
• If the file doesn’t exist, you’ll be prompted to create it.
• Click Apply All to review every code block. Choosing Accept or Reject will automatically move you to the next block.

Provides smart single-line suggestions as you type.

Works like most code editors: as you type, a suggestion popup appears, allowing you to pick from a list of options. Open tabs are automatically used as context to provide more accurate suggestions. Press Shift+Cmd+Space trigger autocomplete suggestions manually.

Shows inline ghost-text suggestions, single or multi-line, to help you complete code quickly.

Press Ctrl + M to trigger Gray Code completion in the editor.

• Press Tab to accept the suggestion.
• Press Esc to dismiss it.

Access this tool from the left-side pane. When the search bar is focused, all available tools will be displayed, and you can search for specific tools. This tool provides Git integration with commit and push functionality. When selected, it automatically generates a commit message based on your Git diffs. You can then push your commits to the remote repository directly from the interface.

This is an experimental feature. To use voice, toggle Caps Lock on to start and off to stop. When you start voice input, the current active chat will be rendered, and the transcribed text will appear in the chat input box. Setup instructions are available in the documentation

Join our Discord server to connect with other users, get support, and discuss features:

Join our Discord