| name | server-doctor |
| description | Public-safe VPS/server diagnostic workflow for SSH access, disk/memory/process checks, Docker/nginx/systemd triage, service health, and deployment readiness. |
| version | 1.0.0 |
| license | MIT |
| metadata | {"hermes":{"tags":["server","vps","ssh","devops","diagnostics","systemd","nginx","docker","workshop"],"related_skills":["public-endpoint-ops","porkbun-api-dns"]}} |
Server Doctor
Use this for server diagnostics and safe operational checks.
Safety rules
- Read-only first: identify host, user, services, ports, disk, memory, logs.
- Never print secrets from
.env, process env, config files, shell history, or logs.
- Do not restart, delete, reset, upgrade, or change firewall/DNS without explicit approval.
- Separate evidence from guesses.
First-pass checklist
hostname && uname -a
whoami && pwd
df -h
free -h
ss -ltnp
systemctl --failed || true
For app servers, run non-mutating checks first. Use elevated privileges only outside this skill after the operator explicitly approves the target and action.
systemctl status <service> --no-pager || true
journalctl -u <service> -n 120 --no-pager || true
docker ps --format 'table {{.Names}} {{.Status}} {{.Ports}}' || true
nginx -t || true
Output contract
Target:
Evidence:
Health:
Likely cause:
Safe next action:
Approval needed for:
Done criteria
The user knows whether the server is healthy, degraded, or blocked, and the next mutation is explicitly scoped before it happens.
