MCP server that lets AI agents drive a real terminal: type, press keys, click, read the screen, take screenshots
Setup
Choose your client, copy the config, then fill in any required environment variables.
claude mcp add terminal-use -- npx -y terminal-use~/.codex/config.toml
[mcp_servers.terminal-use]
command = "npx"
args = ["-y", "terminal-use"]
# codex mcp add terminal-use -- npx -y terminal-use~/.cursor/mcp.json
{
"mcpServers": {
"terminal-use": {
"command": "npx",
"args": [
"-y",
"terminal-use"
]
}
}
}.vscode/mcp.json
{
"servers": {
"terminal-use": {
"type": "stdio",
"command": "npx",
"args": [
"-y",
"terminal-use"
]
}
}
}~/.gemini/settings.json
{
"mcpServers": {
"terminal-use": {
"command": "npx",
"args": [
"-y",
"terminal-use"
]
}
}
}claude_desktop_config.json
{
"mcpServers": {
"terminal-use": {
"command": "npx",
"args": [
"-y",
"terminal-use"
]
}
}
}.trae/mcp.json
{
"mcpServers": {
"terminal-use": {
"command": "npx",
"args": [
"-y",
"terminal-use"
]
}
}
}