ssh-exec

What This Skill Does

The ssh-exec skill is a powerful automation utility designed for OpenClaw AI agents to interact with remote infrastructure seamlessly. It enables the agent to execute non-interactive commands on any reachable Tailscale node. By abstracting the complexities of SSH connections, it allows your AI assistant to act as a system administrator, performing tasks such as checking system health, verifying service statuses, or triggering remote deployment scripts without needing a persistent shell connection.

Installation

To integrate this skill into your OpenClaw environment, execute the following command in your terminal:

clawhub install openclaw/skills/skills/xejrax/ssh-exec

Ensure that your environment has properly configured SSH keys (typically located in ~/.ssh/id_rsa) or an active SSH agent (SSH_AUTH_SOCK) to allow passwordless authentication to the target nodes. Additionally, confirm that the SSH_TARGET environment variable is defined if you intend to use a default target for repeated operations.

Use Cases

This skill is ideal for DevOps automation and remote fleet management. Common scenarios include:

• Remote Health Monitoring: Periodically checking disk usage (df -h) or resource consumption (top -b -n 1) across your Tailscale network.
• Deployment Orchestration: Triggering remote pull requests or restarting system services (systemctl restart service-name) when specific events are detected.
• Audit and Compliance: Running remote security audits, such as listing active processes or checking for unauthorized login attempts.
• Script Dispatch: Pushing local configuration scripts to remote hosts to synchronize state across your infrastructure.

Example Prompts

• "Connect to the production web server and verify if the Nginx service is currently running."
• "Run uptime and free -m on the database node and report the current memory pressure."
• "Execute the cleanup script on the remote storage node located at 100.107.204.64."

Tips & Limitations

This skill is designed for single-command execution. It does not support interactive TTY sessions (like entering a password prompt or using a text editor). For long-running processes, ensure you use nohup or tmux on the remote host. Always verify your SSH identity is added to the target node's ~/.ssh/authorized_keys file before execution. Furthermore, be mindful that the skill executes with the privileges of the user associated with the SSH key; adhere to the principle of least privilege by using limited-scope accounts for your AI agents.