SSH for AI Agents & Automation

Run commands on
your servers — from
anything.

ShellExec gives Claude, Cursor, and any AI agent secure SSH access to your infrastructure via MCP — or hit the REST API directly from curl, Python, or any script. No local setup. No exposed keys.

✓ You're on the list. We'll be in touch shortly.

Free beta · No credit card · 5 minute setup

shellexec — mcp + rest api

# via AI agent (MCP) You: "investigate the high CPU on prod" → ssh_exec: top -bn1 | head -20 → ssh_exec: journalctl -u nginx --since '1h ago' → ssh_exec: systemctl restart nginx Claude: "Fixed. Stale worker at 94% CPU. Service restarted. All normal." # via curl (REST API) $ curl -X POST https://api.shellexec.io/exec \ -H "Authorization: Bearer $TOKEN" \ -d '{"server":"prod","command":"df -h"}' {"stdout":"50G total, 12G used","exit_code":0}

// works with

{
  "mcpServers": {
    "shellexec": {
      "type": "http",
      "url": "https://mcp.shellexec.io/YOUR_TOKEN"
    }
  }
}

~/.config/claude/claude_desktop_config.json

// how it works

Up and running in five minutes.

One URL. One config line. Your AI agent can touch your servers with full audit logs and you in control.

01 —

Add your servers

Paste your SSH key in the dashboard. We store it encrypted. Your agent never sees it.

02 —

Add one config line

Point your MCP client at your ShellExec endpoint. That's the entire setup.

03 —

Tell your agent to act

"Check disk space on prod." "Restart nginx." "Pull the last 50 error logs."

04 —

Every command is logged

Full audit trail. See exactly what ran, when, and what it returned.

// vs self-hosting

Skip the setup.
Keep the control.

Open-source SSH MCP servers exist. But they run locally, need manual setup, and only work on your machine.

Self-hosted

Clone repo, install Node, configure SSH paths
Only works on your machine
No audit logs out of the box
You manage updates when it breaks
Credentials in a local config file

ShellExec

Add one URL to your MCP config — done
Shared across your whole team
Every command logged with full output
We handle updates and uptime
Keys encrypted at rest, never exposed

// features

Built for teams
who trust AI agents.

Multi-server management

Command allowlists

Restrict what commands an agent can run. Read-only on prod, full access on staging.

Full audit log

Every execution recorded — command, output, timestamp, which agent, which server.

Team access

Share your workspace with your team. Everyone gets the same servers and restrictions.

REST API

Not using MCP? Send a curl request with your API key. Same security model.

Key isolation

SSH keys encrypted at rest, injected at runtime. The AI model never sees them.

// pricing

Simple. Per team,
not per seat.

SOLO

Personal use

£9 / month

For individual devs and homelab use.

Up to 5 servers
1,000 executions/month
Full audit log
MCP + REST API

TEAM

Small teams

£29 / month

For teams managing real infrastructure.

Unlimited servers
10,000 executions/month
Command allowlists
Team sharing + roles
Priority support

Be first to try it.

We're building this now. Early access is free. Help us shape what it becomes.