AI Coding Assistant, Powered by Grok.

An agentic AI coding assistant powered by xAI's Grok. Streaming chat, tool calling (file read/write, shell commands, search), code generation, and more. Available as a desktop app, web app, standalone executable, pip package, and a Neovim plugin.

Download the latest release for your platform:

Download from Releases

Requires: Python 3.10+ installed on your system (download). Works with Python 3.10, 3.11, 3.12, 3.13, and newer.

Set your API key in Settings (gear icon) after launching.

pip install tetsuocode

export XAI_API_KEY="xai-..."
tetsuocode              # opens browser to current directory
tetsuocode /path/to/project --port 8080
tetsuocode --no-browser --password secret

No Python required. Download the single-file executable:

Download from Releases

export XAI_API_KEY="xai-..."
./tetsuocode-linux              # or tetsuocode-windows.exe

Install from the marketplace or build from source:

cd vscode
npm install
npx vsce package
# Install the .vsix: Ctrl+Shift+P → "Install from VSIX"

Set your API key in VS Code settings: tetsuocode.apiKey

Features: sidebar chat panel, right-click context menu (Explain, Refactor, Write Tests, Fix Diagnostics), streaming responses, tool calls, Ctrl+Shift+T to toggle, Ctrl+Shift+A to ask.

git clone https://github.com/tetsuo-ai/tetsuo-code.git
cd tetsuo-code
pip install flask requests
export XAI_API_KEY="xai-..."
cd web && python app.py

Open http://localhost:5000.

• Streaming chat with full markdown rendering and syntax highlighting
• Multi-provider support: xAI Grok, OpenAI, Anthropic, Ollama (local)
• Model switching (grok-4-1-fast-reasoning, grok-3-fast, grok-3, grok-3-mini, and more)
• Ghost text inline completions (Tab to accept)
• AI hover tooltips — double-click any code for instant explanation
• Context-aware smart prompt suggestions
• Auto-summarization when context window fills up
• Conversation fork tree — branch and explore alternate paths

• Autonomous tool loop — Grok reads files, writes code, runs commands on its own
• Diff approval flow — review and approve/reject file edits before they apply
• MCP (Model Context Protocol) server support for extending tool capabilities

• Built-in code editor with syntax highlighting overlay
• Multi-cursor editing (Ctrl+D select next, Ctrl+Shift+D select all)
• Git blame gutter and file history viewer
• Real-time linting (Python, JavaScript, JSON)
• Diagnostics bar with error/warning counts

• Integrated streaming terminal
• Test runner with auto-detection (pytest, jest, go test, cargo test)
• Multi-file code review panel with diffs
• Workspace indexing and token-based search
• File watcher with live reload
• Command palette (Ctrl+K) for quick access to everything

• Workspace-scoped file access — can't read/write outside your project
• Dangerous command detection and blocking
• Optional password protection
• All traffic stays local (no telemetry)

• Chat persistence across sessions (localStorage)
• Save and load named conversations
• Code block copy buttons
• Multiple themes
• Mobile responsive
• Keyboard shortcuts: Enter send, Shift+Enter newline, Ctrl+N new chat, Esc cancel, Ctrl+K palette

For Vim users. Adds an AI chat panel directly in your editor.

Requirements: Neovim >= 0.9, curl, an xAI API key

export XAI_API_KEY="xai-..."

lazy.nvim

{
  "tetsuo-ai/tetsuo-code",
  config = function()
    require("tetsuo").setup()
  end,
}

packer.nvim

use {
  "tetsuo-ai/tetsuo-code",
  config = function()
    require("tetsuo").setup()
  end,
}

Grok has access to these tools and uses them autonomously:

The Neovim plugin also exposes get_current_buffer, get_diagnostics, and list_buffers.

tetsuocode [workspace] [options]

Options:
  workspace          Directory to open (default: current directory)
  -p, --port PORT    Port to run on (default: 5000)
  --host HOST        Host to bind to (default: 127.0.0.1)
  --no-browser       Don't auto-open browser
  --password PASS    Set access password
  --api-key KEY      xAI API key (or set XAI_API_KEY env var)
  --version          Show version

require("tetsuo").setup({
  api_key = nil,            -- or set XAI_API_KEY env var
  model = "grok-4-1-fast-reasoning",
  base_url = "https://api.x.ai/v1",
  max_tokens = 4096,
  temperature = 0.7,

  ui = {
    width = 0.38,           -- chat panel width (fraction of editor)
    position = "right",     -- "right" or "left"
    border = "single",
  },

  keymaps = {
    toggle_chat = "<leader>tc",
    ask = "<leader>ta",
    inline_edit = "<leader>ti",
    reset = "<leader>tr",
    fix_diagnostics = "<leader>tf",
  },

  tools = {
    enabled = true,
    max_iterations = 10,
    confirm_writes = true,
    confirm_bash = true,
    bash_timeout = 30000,
  },
})

Create a .tetsuorc in your project root:

{
  "model": "grok-4-1-fast-reasoning",
  "temperature": 0.5,
  "system_prompt": "You are working on a Rust project using Actix-web."
}

pip install flask requests pyinstaller
python build_exe.py
# Output: dist/tetsuocode (or dist/tetsuocode.exe on Windows)

cd electron
npm install
npm run build          # builds for current platform
npm run build:win      # Windows
npm run build:mac      # macOS
npm run build:linux    # Linux

GNU General Public License v3.0