MCP SSH Client

Example Use Cases

• 1

  Execute commands on remote servers via SSH while enforcing automated safety policies to prevent destructive actions.

• 2

  Integrate an LLM-based security filter (via a local Ollama model) to analyze and block unsafe command patterns before execution.

• 3

  An MCP-compatible SSH backend to plug into Claude Desktop, Cursor AI, VSCode Insider, or similar MCP-enabled clients.

Description

This MCP server exposes an SSH client that can be plugged into MCP-capable applications and includes an optional agentic security agent to detect and prevent unsafe commands from executing. The security agent relies on a local Ollama LLM (e.g., llama2) and requires running Ollama locally. It is useful for adding an automated safety layer to remote command execution workflows and has been tested with Claude Desktop, Cursor AI, and VSCode Insider.