FOR AI AGENT DEVELOPERS

Your AI agent needs hands on your servers.
sshDCommander is those hands.

Persistent SSH daemon for Claude Code, Cursor, Windsurf, and any AI agent with shell access. No dropped connections between tool calls. No credentials in prompts. No reconnect overhead.

Try Free — No Credit CardAI Instructions

The problem every AI agent hits

AI coding agents are powerful on your local machine. But the moment they need to touch a remote server — deploy code, check logs, restart a service — they hit the SSH wall.

Every tool call creates a new SSH connection. The connection drops when the tool returns. Next call: reconnect, re-authenticate, lose context. Credentials end up in the prompt or environment variables. Complex commands with pipes and quotes break silently across shell layers.

This is not an AI problem. It is an SSH problem.

How sshDCommander solves it

sshDCommander runs a persistent daemon on your machine. SSH connections stay alive across all tool calls. Your AI agent calls three simple CLI tools.

claude-code — AI agent session
# AI agent executes via sshDCommander CLI — connection persists
$ sshdcmd --server prod --client-id claude "systemctl status nginx" --connect
Connected to 10.0.1.5:22 as deploy
nginx.service - active (running) since Mon 2026-03-28 09:15:01 UTC
$ sshdcp --server prod --client-id claude upload ./release.tar.gz /opt/app/
SHA-256 verified release.tar.gz (24.7 MB) in 1.2s
$ sshdcmd --server prod --client-id claude "cd /opt/app && tar xzf release.tar.gz && systemctl restart app"
app.service restarted successfully
# Connection persists. Next tool call reuses the same session. Zero overhead.

Persistent sessions

SSH connections survive between tool calls. Zero reconnect overhead. The daemon manages the full lifecycle — your agent just calls the CLI.

AI-optimized CLI

--help-ai with usage examples for every command. Structured JSON output. Mandatory client-id for multi-agent traceability.

Zero credential exposure

Server profiles store connection details. OS keyring stores passwords. Nothing ends up in the AI context window or prompt history.

Works with every AI agent

Any AI agent that can execute shell commands works with sshDCommander.

Claude Code

Primary integration target

Cursor

Full CLI compatibility

GitHub Copilot

CLI and IDE integration

Windsurf

Shell command execution

Aider

Open-source CLI agent

Cline

VS Code shell execution

Codex CLI

OpenAI CLI agent

Any AI agent

Anything with shell access

Coming June 2026: MCP Server for native tool-call integration without CLI wrapping.

What AI agents get with sshDCommander

✓
Persistent sessions
Connections survive between tool calls
✓
Verified file transfers
SHA-256 streaming checksums on every upload
✓
Deploy manifests
Cryptographic record of every deployment
✓
Drift detection
Verify server state matches intent
✓
Script mode
Complex commands wrapped safely — no escaping nightmares
✓
Mandatory audit trail
Every request tagged with --client-id
✓
--help-ai
AI-optimized help with usage examples
✓
Certificate auth
Ed25519 CA — no passwords in prompts, ever

Set up in 60 seconds

quick start
$ pip install sshdcommander
$ sshdctl start-daemon
$ sshdctl server add --name prod --host <ip> --user deploy --store
$ sshdcmd --client-id claude "uptime" --server prod --connect

Add the AI instructions to your CLAUDE.md or .cursorrules and your agent is ready.

Give your AI agent reliable SSH access

Base: €99/year — everything your AI agent needs. 14-day free trial, no credit card required.

View PricingDownload v4.0.0