Skip to main content
MCP server CX Codex

mcp-console-automation — Claude Code MCP server: terminal and SSH

Lets an agent drive local consoles and SSH sessions: run commands, send input, read output, spot errors. Useful when terminal work needs automating.

Install into Claude Code · Platform CLI All methods and platforms ↓
claude mcp add --env LOG_LEVEL=<LOG_LEVEL> --transport stdio mcp-console-automation -- npx -y mcp-console-automation
Open on GitHub

What mcp-console-automation does

The server opens up to 50 console sessions at once, covering local shells such as bash, zsh and pwsh plus remote SSH connections. An agent can type text and special keys (Enter, Tab, Ctrl+C), watch output as it arrives, filter it with regex, or search several patterns together. Long jobs run in the background through a priority queue and can be cancelled or paused.

Test tooling is built in: assertions that output contains a value, matches a regex or equals an expected string, exit-code checks, and session snapshots taken before and after an operation. There is also monitoring for CPU, memory, disk and network, alerts for errors and anomalies, and saved profiles for SSH, Docker, WSL and cloud platforms.

Treat this as a privileged terminal rather than a safe documentation connector, because it runs arbitrary commands and opens remote sessions. Keep MCP tool approvals on, use SSH keys or environment-variable references instead of inline secrets, and leave MCP_DEBUG_LOG and MCP_LOG_DIR unset unless you actually need diagnostics. Session persistence ships disabled, since recovery data can hold commands, paths and environment values. console_save_profile refuses inline passwords, private-key material and passphrases — pass passwordEnvVar, privateKeyEnvVar, privateKeyPath or passphraseEnvVar instead.

What it helps with

  • run a command in a local shell and pull its output
  • reach a remote host over SSH and execute commands there
  • catch an error or stack trace in a long-running process
  • assert on exit codes and output content in a test
  • read CPU and memory metrics for one session

Example prompts for the agent

  • Run npm run build in a new console session and show me the last 50 lines
  • Connect with the prod-web profile and check that nginx is still running
  • Start the migrations in the background and watch the output until done appears

How to install mcp-console-automation

Platform CLI

claude mcp add --env LOG_LEVEL=<LOG_LEVEL> --transport stdio mcp-console-automation -- npx -y mcp-console-automation

Replace <…> with your own values

Step by step, with commands for each platform: How to add an MCP server to Claude Code, Codex, Cursor, VS Code, Gemini CLI and more

Questions about mcp-console-automation

How do I install mcp-console-automation in Claude Code?

Run claude mcp add mcp-console-automation -- npx -y mcp-console-automation. To set a log level, use the variant with the environment variable: claude mcp add --env LOG_LEVEL=<LOG_LEVEL> --transport stdio mcp-console-automation -- npx -y mcp-console-automation.

What does mcp-console-automation do?

It gives an agent console sessions to work with: running commands, sending text and special keys, reading and filtering output, background jobs, test assertions and per-session metrics. Both local shells and remote SSH connections are covered.

Is it safe to hand this server to an agent?

It executes arbitrary commands and opens SSH sessions, so treat it like a privileged terminal. Keep tool approvals enabled, keep credentials in environment variables, and leave session persistence off unless you need recovery.

Similar MCP servers

All MCP servers for Agent setup: memory, context, orchestration →

Project sponsors

Companies that keep this analytics open
Gold sponsor

Impressions, clicks and code copies — open statistics. We keep the numbers public for the same reason NeBlask itself is public. Become a sponsor