Works with the AI Agents you already use

    Claude CodeClaude CodeCursorCursorCodex CLICodex CLIMuseMuseDotsDots+21 more

    SSH Remote Ops: Debug, Deploy & Tunnel From Your Agent

    1

    Your agent can write the fix - this skill lets it go apply it on the server, safely.

    $9

    Secure checkout via Stripe

    Money-back guarantee

    Converts to your local currency at checkout

    Security scanned
    SSH Remote Ops: Debug, Deploy & Tunnel From Your Agent

    SSH Remote Ops: Debug, Deploy & Tunnel From Your Agent

    Example session with this skill installed

    My API on prod-web started returning 502s about an hour ago. Figure out why.

    • Read your context and instructions
    • Compiled the ssh remote ops
    > list_credentials
      prod-web  deploy@203.0.113.10
    
    > ssh_exec  prod-web
      curl -sS -o /dev/null -w "%{http_code}\n" http://127.0.0.1/api/health ; systemctl status api --no-pager | head -n 5
      502
      ● api.service - Node API
         Active: activating (auto-restart) (Result: exit-code) since 14:02:11 UTC
    
    > ssh_exec_raw  prod-web  ["journalctl", "-u", "api", "-n", "40", "--no-pager"]
      14:01:58 node[8812]: Error: ENOSPC: no space left on device, write '/var/log/api/app.log'
      14:01:58 systemd[1]: api.service: Main process exited, code=exited, status=1/FAILURE
    
    > ssh_exec  prod-web
      df -hT -x tmpfs -x devtmpfs ; du -xh --max-depth=1 /var/log 2>/dev/null | sort -rh | head -n 5
      /dev/vda1  ext4  40G  40G  0  100% /
      31G  /var/log
      30G  /var/log/api
    

    Diagnosis

    Symptomnginx returns 502 - upstream api.service is crash-looping
    CauseRoot disk is 100 % full; the API dies with ENOSPC on its first log write
    Culprit/var/log/api/app.log - 30 GB, not covered by logrotate

    Proposed fix (waiting for your OK - nothing has been changed yet)

    1. Truncate the oversized log to its last 50k lines: tail -n 50000 app.log > app.log.tmp && mv app.log.tmp app.log
    2. systemctl restart api
    3. Add /etc/logrotate.d/api (daily, rotate 7, compress) to prevent a repeat

    Reply "go" and I'll apply steps 1-2, re-check health, and show you the logrotate file before installing it.

    Connects securely to your tools. The creator never sees your data.

    About this skill

    Your agent can write the fix - this skill lets it go apply it on the server, safely.

    SSH Remote Ops pairs with the open-source ssh-mcp-server (MIT, 15 tools) and teaches your agent when and how to use each one:

    • Run commands with full shell features, or as argv arrays so regexes, quotes and user input never break or inject.
    • Triage incidents with step-by-step runbooks: disk full, service down / crash-looping, high load, 5xx errors, TLS expiry, cron jobs that "don't run".
    • Deploy and sync files - scp for single files, incremental rsync for directories, with the trailing-slash gotcha handled.
    • Drive interactive programs (psql, mysql, htop, prompts) through real PTY sessions.
    • Open SSH tunnels to reach a remote database or admin panel at 127.0.0.1 - and close them when done.

    Built to be safe on production

    • Host keys are pinned on first connection (or up front from a fingerprint you supply); a server that suddenly presents a different identity is refused, not trusted.
    • Servers are referenced by name (prod-web, staging-db). Hosts and key paths are stored once locally (~/ssh.db); key contents are never read into the chat.
    • Read before write: the agent inspects and reports before touching anything.
    • Asks before every change - restarts, config edits, deletions, upgrades - and shows the exact command and target host.
    • Structured errors (SSH_CONNECTION, AUTH_FAILURE, TIMEOUT, ...) with a recovery strategy for each, so the agent doesn't retry blindly.

    What's in the package

    • SKILL.md - rules, tool-selection table, core workflows
    • references/setup.md - install + config for Claude Code, Claude Desktop, Cursor, VS Code Copilot, Codex CLI, Windsurf, Cline, Gemini CLI
    • references/tools.md - every tool, argument, limit and error type
    • references/playbooks.md - 9 production runbooks

    Requires: Node.js 18+, the free ssh-mcp-server MCP server, and SSH key access to your servers.

    Requires the free ssh-mcp-server MCP server to be installed. Key-based authentication only (no password logins), and servers must listen on the default SSH port 22. Remote commands run through bash, so targets are Linux/Unix hosts. Private keys are referenced by local file path; pasting key contents is not supported. If a server's host key changes, connections are refused until you verify the new key.

    How to use

    1. 1

      Connect your agent to the Agensi MCP

      Copy the prompt and paste it into Claude, ChatGPT, Cursor or any agent that supports MCP. It sets up the connection for you. Manual setup

    2. 2

      Sign in to Agensi

      Your agent shows a sign-in link. Sign in and click Allow.

    3. 3

      Ask for what you need

      Your agent loads this skill and uses it on your task. It always asks before any payment.

    Reviews

    No reviews yet

    Be one of the first to try it. Every listed skill goes through the trust checks below.

    Security scanned

    Checked by our 8-point scan before listing

    Fresh listing

    Recently published to Agensi

    Money-back guarantee

    Doesn't do what its page promises? You get your money back.

    Trust & safety

    Security scan

    Security scan passed 20 days ago

    Requires

    Ssh MCP
    Node.js 18+
    rsync (optional)
    Git
    Node.js
    @smithery/cli

    Permissions

    Terminal / Shell
    Network Access
    Read Files
    Write Files

    File Scopes

    ~/ssh.db

    Terminal/Network: the skill drives the ssh-mcp-server tools, which open SSH connections to servers the user has registered. Read/Write Files: file transfers (scp/rsync) read and write only the local paths the user chooses; credentials live in ~/ssh.db and only the private key path is stored, never key contents. The skill contains no scripts and makes no other network calls.

    Listed20 days ago

    What's inside

    Frequently Asked Questions

    Yes - the free, open-source ssh-mcp-server (MIT) and Node.js 18+. references/setup.md in the package has copy-paste config for Claude Code, Claude Desktop, Cursor, VS Code Copilot, Codex CLI, Windsurf, Cline and Gemini CLI. The skill is the operating knowledge: when to use which tool, safety rules and runbooks.

    The skill instructs the agent to inspect first and to ask for explicit confirmation before any change - restarts, config edits, deletions, upgrades, firewall changes - showing the exact command and the target server. Read-only diagnostics run without prompting.

    You register each server once under a name. Only the host, username and the path to your private key are stored, locally in ~/ssh.db - key contents are never read into the chat or sent anywhere. Each server's host key is pinned on first connection (or from a fingerprint you supply), and a changed host key is refused.

    Linux and other Unix hosts reachable over SSH on port 22 with key-based login and bash available. The runbooks assume common tooling (systemd, journalctl, nginx, Docker) but the agent adapts when something is missing. Password-only logins are not supported.

    Yes, through PTY sessions: the agent can drive psql, mysql, htop or any prompt. When a password is requested, the agent stops and lets you decide - it never invents or reuses one.

    New in DevOps & Cloud