sideshell
AI sidecar terminal — let Claude/Cursor run commands in a visible, persistent terminal you control.
README
sideshell
AI sidecar terminal — let any MCP client (Claude Code, Cursor, Codex, OpenCode, Pi, …) run commands in a visible, persistent terminal you control.
Why?
When AI assistants run shell commands, they execute in a hidden terminal:
- ❌ No visible history
- ❌ Can't intervene (enter password, confirm prompts)
- ❌ Output mixed with AI conversation
sideshell runs commands in a separate visible terminal:
- ✅ Full command history visible
- ✅ Persistent session (survives AI restarts)
- ✅ Intervene anytime (passwords, confirmations)
- ✅ Clean separation from AI conversation
Supported Terminals
| Terminal | Platform | Status |
|---|---|---|
| iTerm2 | macOS | ✅ Full support (native Python API) |
| tmux | macOS, Linux, WSL | ✅ Full support |
| WezTerm | macOS, Linux, Windows | ✅ Full support |
| Kitty | macOS, Linux | ✅ Full support |
| Ghostty | macOS | ✅ Full support (ghostty_tmux hybrid: native AppleScript splits + per-surface tmux engine, Ghostty 1.3+) |
| maquake | macOS | ✅ Full support (drop-down terminal via Unix socket) |
| VS Code / Cursor | macOS, Linux, Windows | ✅ Full support (via extension, Unix-socket bridge) |
| JetBrains IDEs | macOS, Linux, Windows | ✅ Full support (via plugin, Unix-socket bridge) |
Features
- Multi-Backend - Works with iTerm2, tmux, WezTerm, Kitty, Ghostty, maquake, VS Code/Cursor, or JetBrains IDEs
- Sidecar Terminal - AI commands run in a visible terminal pane
- You Stay in Control - See everything, intervene anytime
- Session Persistence - Terminal survives AI session restarts
- TUI Support - Arrow keys, F1-F12, Ctrl+C/D/Z for interactive apps
- Focus Management - Optionally returns focus after operations
Installation
Using uvx (Recommended)
# Install uv first
curl -LsSf https://astral.sh/uv/install.sh | sh
# Run sideshell
uvx sideshell-mcp
Claude Desktop (~/Library/Application Support/Claude/claude_desktop_config.json):
{
"mcpServers": {
"sideshell": {
"command": "uvx",
"args": ["sideshell-mcp"]
}
}
}
Using pipx
pipx install sideshell-mcp
sideshell-mcp
Backend Selection
# Auto-detect (default)
uvx sideshell-mcp
# Force specific backend
uvx sideshell-mcp --backend=tmux
uvx sideshell-mcp --backend=iterm2
Available Tools (17)
| Tool | Description |
|---|---|
execute |
Execute commands (supports wait, timeout, targets for broadcast) |
read |
Read terminal output |
control-char |
Send special keys: Ctrl+C/D/Z, arrows, F1-F12, Home/End, PageUp/Down |
list |
List all windows/tabs/sessions |
split |
Split pane horizontally or vertically |
new-window |
Create new window |
new-tab |
Create new tab |
new-session |
Smart session creation (splits if window exists) |
focus |
Focus specific session |
close-session |
Close terminal session |
set-appearance |
Set tab title, badge, and color |
get-terminal-state |
Get detailed terminal state |
list-color-presets |
List available color presets |
set-color-preset |
Apply color preset |
show-alert |
Show alert dialog |
clear |
Clear terminal screen |
paste |
Paste text to terminal |
MCP Resources
| Resource | Description |
|---|---|
sideshell://sessions |
List all terminal sessions |
sideshell://capabilities |
Backend features and system info |
sideshell://sessions/{id} |
Session details |
sideshell://sessions/{id}/screen |
Screen content |
Prerequisites
iTerm2 (macOS)
- Open iTerm2 → Preferences → General → Magic
- Enable "Enable Python API"
- Restart iTerm2
tmux
# macOS
brew install tmux
# Ubuntu/Debian
sudo apt install tmux
WezTerm
Download from wezfurlong.org/wezterm
Kitty
# macOS
brew install --cask kitty
# Linux
curl -L https://sw.kovidgoyal.net/kitty/installer.sh | sh
Architecture
┌─────────────┐ ┌──────────────┐ ┌─────────────┐
│ MCP Client │────▶│ sideshell │────▶│ Terminal │
│ (Claude) │ │ MCP Server │ │ Backend │
└─────────────┘ └──────────────┘ └─────────────┘
│
┌───────┴────────────┐
│ Backends │
├────────────────────┤
│ • iTerm2 │
│ • tmux │
│ • WezTerm │
│ • Kitty │
│ • Ghostty │
│ • maquake │
│ • VS Code / Cursor │
│ • JetBrains IDEs │
└────────────────────┘
VS Code/Cursor and JetBrains backends talk to their IDE extension/plugin over a
local Unix domain socket (~/.sideshell/<ide>.sock) using newline-delimited
JSON-RPC 2.0 with a token handshake.
Development
git clone https://github.com/menemy/sideshell
cd sideshell
uv pip install -e ".[dev]"
# Run tests
python tests/test_iterm2_backend.py # iTerm2
python tests/test_tmux_backend.py # tmux
python tests/test_wezterm_backend.py # WezTerm
python tests/test_kitty_backend.py # Kitty
# Lint & format
ruff format .
ruff check . --fix
Requirements
- Python 3.11+
- One of: iTerm2, tmux, WezTerm, Kitty, Ghostty, maquake, VS Code/Cursor, or a JetBrains IDE
License
MIT
Contributing
- Fork the repository
- Create a feature branch
- Add tests for new functionality
- Ensure all tests pass
- Submit a pull request
Recommended Servers
playwright-mcp
A Model Context Protocol server that enables LLMs to interact with web pages through structured accessibility snapshots without requiring vision models or screenshots.
Magic Component Platform (MCP)
An AI-powered tool that generates modern UI components from natural language descriptions, integrating with popular IDEs to streamline UI development workflow.
Audiense Insights MCP Server
Enables interaction with Audiense Insights accounts via the Model Context Protocol, facilitating the extraction and analysis of marketing insights and audience data including demographics, behavior, and influencer engagement.
VeyraX MCP
Single MCP tool to connect all your favorite tools: Gmail, Calendar and 40 more.
graphlit-mcp-server
The Model Context Protocol (MCP) Server enables integration between MCP clients and the Graphlit service. Ingest anything from Slack to Gmail to podcast feeds, in addition to web crawling, into a Graphlit project - and then retrieve relevant contents from the MCP client.
Kagi MCP Server
An MCP server that integrates Kagi search capabilities with Claude AI, enabling Claude to perform real-time web searches when answering questions that require up-to-date information.
E2B
Using MCP to run code via e2b.
Neon Database
MCP server for interacting with Neon Management API and databases
Exa Search
A Model Context Protocol (MCP) server lets AI assistants like Claude use the Exa AI Search API for web searches. This setup allows AI models to get real-time web information in a safe and controlled way.
Qdrant Server
This repository is an example of how to create a MCP server for Qdrant, a vector search engine.