homelabhero

Top-level triage and health-overview workflow for the homelab. Use this whenever Evan asks broad or symptom-first questions like "what's wrong", "is everything okay", "something's down", "the homelab feels slow", "give me an overview", "health check", "status of everything", or reports a problem without naming the layer. Also use it as the first step for any incident where the failing component is not yet obvious, before dropping into the host-specific skills. Reach for this even when Evan does not say the word "triage" but is clearly asking for a whole-system look or reporting a vague fault.

Troubleshoot and manage TrueNAS SCALE over SSH. Use this whenever Evan mentions TrueNAS, the NAS, ZFS, a pool or dataset, a scrub, a snapshot, replication, a failing or degraded disk, SMART data, an SMB/NFS/iSCSI share, a TrueNAS app, or storage that is full, slow, or throwing errors. Trigger this for anything storage-layer even when the symptom first shows up as an app or VM that lost its data mount. Prefer this over generic shell reasoning for pool, disk, dataset, and share work.

Troubleshoot and manage the Docker / Dockge stacks over SSH, especially the arr media stack. Use this whenever Evan mentions a container, Docker, Dockge, a compose stack, or any specific app such as Jellyfin, qBittorrent, gluetun, Radarr, Sonarr, Prowlarr, Seerr, Immich, Seafile, SearXNG, or similar, and whenever the symptom is app-level: a container is down, restarting, unhealthy, downloads are stalled, requests are not processing, or a service is unreachable while its host is fine. Trigger this even if he just names an app and says it is broken.

Register a new server so HomelabHero can manage it, driven entirely from the chat/UI with no shell access and no password ever entering the conversation. Use this whenever Evan (or a user) asks to add, register, connect, onboard, or "hook up" a server, host, machine, NAS, node, or box, or says things like "add my truenas", "connect to my proxmox at 10.x.x.x", or "I have another Linux box to manage". Trigger it any time someone wants a new machine brought under management, even if they do not say the word "register".

Understand and operate the homelab's snapshots and backups: how they work on each platform and how to take, list, verify, and restore/roll back safely. Use this whenever Evan asks to snapshot, back up, restore, roll back, or recover a VM, LXC, dataset, pool, config, or app; asks "can I undo this", "is there a backup", "how do snapshots work here", or "are the backups current"; or right before you propose a risky/destructive change and want to offer a snapshot as a safety net. Do NOT snapshot everything pre-emptively on every task - reach for this when recovery is the actual subject, or to offer one safety net before a specific dangerous operation.

Stand up a NEW containerized app or compose stack on a docker host. Use whenever Evan wants to deploy, install, add, spin up, or "set up" a new container or stack (Jellyfin, Radarr, Sonarr, qBittorrent, Immich, a new arr app, a dashboard, etc.), or add a service to an existing stack. For fixing or operating an app that already exists, use docker-stack-ops instead; for updating images, use patch-management.

Troubleshoot and manage generic Linux hosts (anything that is not Proxmox or TrueNAS). Use this whenever Evan points at a plain Linux server, VM, or container host by alias and asks about services, disk space, networking, updates, processes, or logs, or when a host is slow, down, or misbehaving. It drives systemd, packages, storage, and networking over SSH. Trigger this for any host-level work the Proxmox and TrueNAS skills do not cover.

Diagnose homelab connectivity, DNS, mesh, and ingress issues. Use this whenever Evan reports that hosts or services are unreachable, DNS is not resolving, the NetBird mesh is flaky, a Cloudflare Tunnel is down, the UCG-Ultra gateway or Netgear 10GbE switch is involved, a link dropped or renegotiated speed, or a service works locally but not remotely. Trigger this early in any "everything is down" situation, since broad outages are usually a network or DNS layer wearing an application costume.

Update and patch the managed hosts and their containers safely. Use whenever Evan asks to update, upgrade, or patch a host, VM, node, or app; asks "are we patched", "any updates", "update proxmox/truenas/ubuntu", "update my containers", or wants to apply pending OS or image updates. Covers Debian/Ubuntu hosts, Proxmox, TrueNAS, and Docker containers. Diagnose what's pending first, snapshot before applying, confirm, then verify. This is about the MANAGED hosts, not the HomelabHero control plane.

Troubleshoot and manage Proxmox VE over SSH. Use this whenever Evan mentions Proxmox, PVE, a VM or LXC container, a hypervisor node, "qm", "pct", cluster quorum, HA, migration, a VM that won't boot or is stuck, a node running hot or unreachable, VM/container backups (vzdump), or Proxmox storage. Trigger this even if he only names a VM or container by ID or purpose and wants it started, stopped, checked, or fixed. Prefer this over generic shell reasoning for anything hypervisor-level.

Read-only security-posture review of a managed host or the whole estate: pending updates, listening services and open ports, SSH hardening, user accounts and sudo, failed logins, firewall state, and platform-specific exposure. Use whenever Evan asks to audit, harden, or security-check a box, or asks "is this exposed", "what ports are open", "any security issues", "are we patched", or wants a hardening review. Diagnose and report first; propose fixes and get an explicit go-ahead before changing anything.

Discover and use the live TrueNAS middleware method surface instead of guessing. Use this whenever a TrueNAS task needs a midclt method that is not in capabilities/truenas.md, whenever Evan asks about the TrueNAS API, the middleware, midclt, "what can the API do", available methods, or method parameters, or whenever a midclt call fails with an unknown-method or bad- argument error. Trigger this to look up the real methods, their exact names, and their parameter schemas on THIS box rather than relying on the static catalog. Prefer this over guessing method names for anything beyond the common operations already documented.

Build a complete live inventory of everything running across the homelab. Use this whenever Evan asks what is running, what VMs / LXCs / containers / apps exist, "what's on <host>", "give me the full picture", "what do I have", capacity of guests, or wants an audit of the whole estate. Also use it as the grounding step before any change so you are working from current truth, not assumptions. Trigger it even when he does not say the word "inventory" but is clearly asking what exists or what is deployed where.