NEKOT

A powerful and intuitive terminal utility for interacting with both local and cloud LLMs.

A tool for those who appreciate keyboard driven apps and terminal workflows (mouse works too btw)

Features

• 📦 Single binary - lightweight, zero dependencies, use anywhere without any requirements
• 🤖 Support for OpenAI compatible APIs (ChatGPT, Mistral, Ollama, LMStudio, llama-cpp and more)
• 🌟 Support for Gemini API
• 🔀 Support for OpenRouter API
• 🖼️ Images support
• 🔍 Web search tool [BETA]
• 💬 Chat sessions management and quick chats
• ⚙️ Settings presets (configure different personas with unique settings)
• ✂️ Convenient text selection tool (vim-like line selection)
• 🌐 Crossplatform - support for MacOS, Windows and Linux
• 🎨 Multiple themes

Demo

App installation

Ensure API keys are configured before running NeKot

A good terminal with GPU acceleration is recommended (the app is mostly tested on Ghostty)

Homebrew (MacOS, Linux)

brew tap BalanceBalls/BalanceBalls
brew install nekot --cask

Chocolatey (Windows)

choco install nekot

Shell script (Linux, MacOS)

Requires curl and tar.

For updating the app, just run the script again

Basic install:

• Requires sudo, and installs to /usr/local/bin/

curl -fsSL https://raw.githubusercontent.com/BalanceBalls/nekot/main/install.sh | sudo sh

Install to path:

• Use -p flag to specify install path.
• Make sure the specified directory is in the $PATH.
• May require sudo depending on the path specified.

bash -c "$(curl -fsSL https://raw.githubusercontent.com/BalanceBalls/nekot/main/install.sh)" -- -p ~/.local/bin

Download binary/executable (Windows, MacOS, Linux)

1. Go to latest release
2. Download and extract the archive
3. Place the binary/executable in some directory
4. Add that directory to $PATH

Setting API keys

NeKot can load the active provider's API key from a command or from its existing environment variable. To use a secure store, add an apiKeyResolveCommand command to ~/.nekot/config.json:

{
  "apiKeyResolveCommand": "cmd:security find-generic-password -s openrouter_nekot_key -w"
}

Everything after cmd: is executed once at startup using /bin/sh -c on macOS and Linux, or cmd /C on Windows. The command's trimmed standard output is used as the API key. This also supports password-manager CLIs, for example:

{
  "apiKeyResolveCommand": "cmd:bw get password openrouter_nekot_key"
}

The command inherits NeKot's environment, including variables such as BW_SESSION. Do not put secrets directly in the command. If apiKeyResolveCommand is absent or empty, NeKot falls back to OPENAI_API_KEY, GEMINI_API_KEY, or OPENROUTER_API_KEY, depending on the active provider. A configured command takes precedence and a command failure does not fall back to the environment. Plaintext values in apiKeyResolveCommand are rejected. Treat config files that contain apiKeyResolveCommand like scripts: do not copy or run configs from people you do not trust.

export OPENAI_API_KEY="some-key" # you would want to export this in your .zshrc or .bashrc

export  GEMINI_API_KEY="some-key" # you would want to export this in your .zshrc or .bashrc

export  OPENROUTER_API_KEY="some-key" # you would want to export this in your .zshrc or .bashrc

Stdin piping

You can pipe input into the app and do things like this:

cat PROMPT.md | nekot -p openai -u http://localhost:11434 -m gpt-oss:latest -n

Web search (BETA)

Web search feature uses tool calling for web search reqests exectution. This feature can be toggled using Ctrl+w.

Details

• Uses the DuckDuckGo and Brave search engines and requires no configuration
• Exposes a current_datetime tool alongside web_search so models can check the local date, time, and timezone before current-day searches
• Results are scored using bm25 for better accuracy
• Using web search can significantly increase token usage

Config

We provide a config.json file within your directory for easy access to essential settings.

• On MacOS & Linux, the path is ~/.nekot/config.json.
• On Windows, the path is C:\Users\%UserName%\.nekot\config.json or %HOMEPATH%\.nekot\config.json

Example

{
  "providerBaseUrl": "https://api.openai.com", // Or http://localhost:11434, or any other OpenAi compatible API
  "apiKeyResolveCommand": "", // Optional cmd: command; falls back to the provider's environment variable
  "systemMessage": "",
  "defaultModel": "",
  "colorScheme": "groove", // pink, blue, groove
  "provider": "openai", // openai, gemini, openrouter
  "maxAttachmentSizeMb": 3,
  "includeReasoningTokensInContext": true,
  "sessionExportDir": "/must/be/absolute/path/to/exports",
  "titleGeneration": {
    "enabled": true,
    "timeoutSeconds": 5
  }
}

• providerBaseUrl: The url can be anything that follows OpenAI API standard ( ollama, lmstudio, etc)
• apiKeyResolveCommand: optional cmd: command whose output provides the active provider's API key
• chatGPTApiUrl [obsolete]: same as providerBaseUrl
• systemMessage field is available for customizing system prompt messages. Better to set it from the app
• defaultModel field sets the default model. Better to set it from the app
• maxAttachmentSizeMb field sets maximum allowed image size
• includeReasoningTokensInContext field sets whether to include reasoning tokens in the next request or not.
• sessionExportDir allows to specify directory for session exports. If not set, exports are saved to current directory. The path must be an absolute path
• titleGeneration config for automatic session name generation

Providers

You can change API provider using the provider field.

Available providers:

• openai default
• gemini
• openrouter

To use GeminiAPI, set "provider": "gemini" and configure apiKeyResolveCommand or GEMINI_API_KEY. When using the gemini or openrouter providers, providerBaseUrl param is not used.

Themes

You can change colorscheme using the colorScheme field.

Available themes:

• groove default
• pink
• blue

Data migration

If you need your settings and chats on other machine - simply copy chat.db and config.json from the .nekot/ directory and paste to the same directory on other machine.

Flags

Cache invalidation

Models list is cached for 14 days upon loading. If you need to invalidate cache use --purge-cache flag:

nekot --purge-cache

Provider

To switch between openai and gemini APIs you can use -p flag:

nekot -p openai
nekot -p gemini
nekot -p openrouter

Provider url

To specify provier url use -u flag(works with the openai provider only):

nekot -u http://localhost:11434

Theme

To specify color scheme use -t flag:

nekot -t blue

Other flags

To specify a model use -m flag:

nekot -m "qwen3:4b"

To create a new session on start use -n flag:

nekot -n

Global Keybindings

• Tab: Change focus between panes. The currently focused pane will be highlighted
• 1-4 pane jumps: 1 prompt pane, 2, chat pane, 3 settings pane, 4 sessions pane
• Ctrl+b or Ctrl+s: Interrupt inference
• Ctrl+o: Toggles zen mode
• Ctrl+c: Exit the program
• Ctrl+n: Create new session
• Ctrl+w: Toggles web search (preset level setting)
• Ctrl+h: Hide/show reasoning tokens (preset level setting)
• Ctrl+q: Start quick chat
• Ctrl+x: Save quick chat to session

Prompt Pane

• i: Enters insert mode (you can now safely paste messages into the tui)

• Ctrl+e: Open/Close prompt editor

• Ctrl+r: Clear prompt

• Ctrl+v: Paste text from the clipboard, or attach an image copied to the clipboard

• Ctrl+s: Paste text from buffer as a code block (only in editor mode)

  • if current line contains text, that text will be used as a language for the code block
  • Example: if a line contains go the result of Ctrl+s will be:

  ```go
  {bufferContent}
  ```

• esc: Exit insert mode for the prompt

  • When in 'Prompt editor' mode, pressing esc second time will close editor

• Ctrl+a: open file picker for attaching images. You can also attach images by typing: [img=/path/to/image]

Chat Messages Pane

• y: Copies the last message into your clipboard.
• Shift+y: Copies all messages from current session into your clipboard.
• v, Shift+v or space: Enters navigation mode when chat pane is focused (allows to move accross the chat content lines)

Selection mode

Mouse controls

• Left mouse button to enter line selection mode. Click left mouse button to enter selection mode, then hold left mouse button to select lines

• Right mouse button to enter character selection mode (within a line). Click right mouse button to enter selection mode, then hold right mouse button to select text

Keyboard controls

Selection mode allows to navigate the chat pane and select lines to copy. Supports basic vim-motions.

Navigation

• j, k - go down and up a line
• d, u, Ctrl+d, Ctrl+u - go up or down half page
• g - go to top
• Shift+g - go to bottom

Selection

• d, u, Ctrl+d, Ctrl+u - go up or down half page
• j, k - go down and up a line
  • Multiline jumps like 3j (3 lines down), 99k (99 lines up) are also supported
• v, Shift+v or space to enter or quit line selection mode
• y to copy selected text (with formatting from the app)
• r, c to copy selected text as raw LLM output
• Esc to quit selection or navigation modes

Settings Pane

• [ and ]: switch between presets and settings tabs

Settings tab

• m: Opens a model picker to change the model. (use / to set filter)
• f: Change the frequency value
• t: Change the maximum number of tokens per message
• e: Change the temperature value
• p: Change the top_p value (nucleus sampling)
• s: Opens a text editor to edit system prompt
• Ctrl+r: resets current settings preset to default values
• Ctrl+p: creates new preset with a specified name from the current preset

Presets tab

• d: remove preset (default and current selected presets cannot be removed)
• enter: select preset as the current one
• /: filter presets

Sessions Pane

• Ctrl+n: Creates a new session.
• Shift+X: Exports session to a markdown file.
• d: Deletes the currently selected session from the list.
• e: Edit session name
• Enter: Switches to the session that is currently selected.
• /: filter sessions

Info pane

Information pane displays processing state of inference (IDLE, PROCESSING) as well as token stats for the current session:

• IN: shows the total amount of input tokens LLM consumed per session
• OUT: shows the total amount of output tokens LLM produced per session

Please refer to this guide as you navigate the TUI. Happy exploring!

Dev notes

The SQL db is stored in you your/home/directory/.nekot, as well as the debug log. To enable debug mode, export DEBUG=1 before running the program.

To get access to the release candidates, install command:

brew install rc-nekot
rc-nekot

Technologies

• Go
• bubbletea: A Go framework for terminal user interfaces. It's a great framework that makes it easy to create TUIs in Go.
• openai: OpenAI's REST Api
• sqlite: CGO-free sqlite lib
• lipgloss: Style definitions for nice terminal layouts!
• bubbles: Some general use components for Bubble Tea apps!
• operouter-sdk openrouter api sdk

Contributors

BalanceBalls 📖 💻

Taranveer (Taran) Bains 📖 🚧 👀 💻

TJ Miller 📖 💻