Aiterm Mcp
Give your coding agent one durable shell session over MCP so long-running servers, SSH hops, and nested docker execs do not vanish between tool calls.
Overview
aiterm-mcp is a MCP server for the Build phase that gives coding agents one persistent tmux-backed terminal over stdio, including nested SSH and container shells.
What is this MCP server?
- Single persistent tmux session the model can read and write across many MCP tool calls
- stdio npm package aiterm-mcp@0.4.1 for standard MCP host wiring
- SSH and container workflows nest inside the same terminal instead of one-shot exec
- Designed for agents that need shell state (cwd, env, background jobs) to survive
- npm package aiterm-mcp version 0.4.1
- stdio MCP transport
- GitHub repository kitepon-rgb/aiterm-mcp
Community signal: 1 GitHub stars.
What problem does it solve?
Agents lose shell context when every command is a fresh exec, which breaks multi-step deploys, SSH chains, and anything that needs a running process in the background.
Who is it for?
Indie builders who let agents run real dev and ops commands locally or over SSH and need session persistence beyond single-shot terminal tools.
Skip if: Teams that only need read-only docs lookup or who forbid agents from shell access on the dev machine.
What do I get? / Deliverables
After you register aiterm-mcp, your agent can drive a stable terminal session so builds, tunnels, and container work continue across tool calls without re-explaining state.
- Registered MCP server exposing a persistent tmux terminal
- Agent-accessible shell suitable for SSH and container nesting
Recommended MCP Servers
Journey fit
Agent tooling is the canonical shelf because the server’s core value is extending what Claude Code, Cursor, and Codex can run locally—not a one-off launch or growth tactic. Persistent tmux-backed terminals are invoked from the agent loop during implementation and debugging, which maps directly to agent-tooling under Build.
How it compares
MCP terminal integration with tmux persistence, not a browser automation or cloud IDE product.
Common Questions / FAQ
Who is aiterm-mcp for?
Solo and small-team builders using Claude Code, Cursor, Codex, or other MCP hosts who want agents to operate a real, stateful shell safely on their machine.
When should I use aiterm-mcp?
Use it when workflows span multiple commands, background processes, SSH jumps, or docker exec and you do not want the agent to lose cwd, env, or job state between turns.
How do I add aiterm-mcp to my agent?
Add the npm stdio server aiterm-mcp (v0.4.1) to your host’s MCP config—typically npx aiterm-mcp—then restart the client so tools can attach to the tmux session.