interminal

What is interminal?

Lightweight MCP server that gives AI assistants terminal access — SSH and local shells — with support for interactive and long-running commands.

How to use interminal?

Install via pip install mcp-interminal or run directly with uvx mcp-interminal. Configure your MCP client (e.g., Claude Desktop) using the command uvx with arguments ["mcp-interminal"]. Requires Python ≥ 3.11.

Key features of interminal

• SSH and local shell execution
• Interactive and long-running command support
• Tools: connect_ssh, execute, read_output, respond, send_control, disconnect
• Optional persistent state with Zellij terminal multiplexer
• Send control keys like ctrl+c, arrow keys, and F-keys
• Optional Windows PTY and ANSI rendering dependencies

Use cases of interminal

• AI-driven remote server management via SSH
• Executing shell commands and chaining with &&
• Multi-step workflows like project development and deployment
• Interactive applications requiring prompt responses
• Monitoring long-running commands with partial output

FAQ from interminal

How does execute handle persistent state?

Each execute call runs in an isolated channel; cd does not persist. For multi-step workflows, install Zellij on the host to provide a persistent session that carries over environment and state.

What are the optional dependencies?

Windows PTY support: pip install mcp-interminal[pty]. ANSI escape rendering: pip install mcp-interminal[ansi]. Both can be installed with pip install mcp-interminal[pty,ansi].

How do long-running commands work?

They return status="partial" with a command_id. Poll for output with read_output or send input with respond while the command runs.

What runtime is required?

Python 3.11 or higher.

How are SSH connections authenticated?

The README does not detail SSH authentication; SSH credentials are provided by the user when connecting. The server communicates with the MCP client via stdio.