Hop
Let your agent run SSH-backed server operations through Hop’s connection manager instead of copying shell one-liners from a password doc.
Overview
Hop is a MCP server for the Operate phase that combines SSH connection management with AI-assisted remote server operations.
What is this MCP server?
- Hop bundles an SSH connection manager with an embedded MCP server for AI-assisted ops
- Version 1.1.0 server manifest from danmartuszewski/hop on GitHub
- Centralizes multi-server access so agents target named hosts instead of ad hoc IPs
- Suited to indie deployments on VPS, bare metal, or small Kubernetes nodes via SSH
- Keeps human-oriented Hop UX while exposing machine-oriented MCP tools
- Hop MCP server schema version 1.1.0
- Repository: github.com/danmartuszewski/hop
Community signal: 53 GitHub stars.
What problem does it solve?
Jumping between terminals, SSH configs, and your agent wastes time when production fixes need both context and shell access.
Who is it for?
Solo builders running one to many Linux servers who want MCP-driven SSH from Hop’s managed connection list.
Skip if: Teams standardized on cloud-only consoles with no SSH, or environments where unrestricted agent shell access violates policy.
What do I get? / Deliverables
Your agent issues operations against hosts you already configured in Hop, reducing friction for deploy checks and infra troubleshooting.
- Agent-executed remote diagnostics and maintenance via Hop
- Consolidated SSH target list reused across terminal and MCP
- Faster incident response without manual host lookup
Recommended MCP Servers
Journey fit
Production and staging servers are operated and debugged long after the first feature ships, which maps to operate/infra. SSH session management is core infrastructure work: connect, diagnose, and change remote hosts safely.
How it compares
SSH connection manager plus MCP, not a full PaaS deploy pipeline or Kubernetes control plane skill.
Common Questions / FAQ
Who is Hop for?
Indie developers and small teams who SSH into their own servers and want AI agents to help with routine remote operations via MCP.
When should I use Hop?
Use it at ship and operate time when you need agent-assisted checks, log inspection, or fixes on remote machines you manage in Hop.
How do I add Hop to my agent?
Install and configure Hop from the GitHub repository, define SSH hosts in Hop, then enable its MCP server in your Claude Code or Cursor MCP settings per Hop docs.