ubuntu-admin

Plugin ubuntu-admin in danielrosehill/Ubuntu-Admin. Commands read from the repo's manifest — 6 days ago.

Skills (49)

• new-workspace

  Provision a new desktop-manager workspace on disk. Use when the user wants a persistent scaffold for managing their local Linux desktop (tracking installs, config changes, hardware issues, log reviews). Accepts a workspace name and optional target parent path. Scaffolds the workspace, personalises C

• onboard

  First-run onboarding for the desktop-manager plugin. Auto-profiles the local Linux desktop (distro, kernel, desktop environment, session type, package managers, audio stack, GPU, hardware summary), then interviews the user for the small handful of things that can't be auto-detected (sudo policy, GUI

• update-profile

  Surgical edits to the desktop-manager machine profile without re-running full onboarding. Use when a single field changed — e.g. user upgraded distro, switched from X11 to Wayland, swapped GPU, changed sudo policy, added a service to watch, recorded a new hardware quirk. Reads and writes `$CLAUDE_US

• desktop-codecs

  Check and complete multimedia codec support on an Ubuntu desktop — GStreamer plugin sets, FFmpeg codecs, DVD playback, and hardware video decode/encode (VA-API, VDPAU, NVDEC) per GPU vendor — and install what is missing (ubuntu-restricted-extras and friends). Use when asked "what codecs do I have",

• desktop-displays

  Inspect and configure monitors on an Ubuntu desktop — list connected displays and modes, detect Wayland vs X11, arrange multiple monitors, set resolution, refresh rate and (fractional) scaling, and save/restore named layouts such as docked, laptop-only and presentation. Covers GNOME (Ubuntu default)

• desktop-fonts

  List, install and manage fonts on Ubuntu — count and search installed families, install from apt (fonts-* packages, Microsoft core fonts), from Google Fonts by name, or from local font files, per-user or system-wide, refresh the cache, and fix basic rendering. Use when asked "what fonts do I have",

• dev-environment

  Inventory and set up a developer environment on Ubuntu — Python (system python3 vs uv, pipx, pyenv, venv and conda detection; never pip into the system interpreter), Node.js (nvm/fnm vs apt/snap, version check), Java and other SDKs via SDKMAN, other toolchains on PATH, and git global configuration (

• docker-cleanup

  Reclaim disk space on a Docker host — report what's using space, then prune selectively (dangling and unused images, stopped containers, unused networks, build cache), trim oversized container json logs, find abandoned compose projects, and remove orphaned volumes only with explicit per-volume confi

• docker-compose-deploy

  Deploy, update and roll back Docker Compose stacks on an Ubuntu host — list stacks, keep one directory (optionally one git repo) per stack, validate with docker compose config, pull, bring up, verify health, and roll back to the previous image or git revision; handle .env files and secrets without c

• docker-health

  Health check and troubleshooting for a Docker host on Ubuntu — daemon state, container status (unhealthy, restarting, exited, dead), restart counts, live resource use, recent events, per-container logs, networks and port bindings, volume mounts and permission/UID problems, and disk use. Read-only by

• docker-setup

  Check for and install Docker on Ubuntu the right way — Docker Engine plus the compose and buildx plugins from Docker's official apt repository (keyring + deb822 source), with the trade-offs against Ubuntu's docker.io package and the snap; add a user to the docker group; configure /etc/docker/daemon.

• files-organise

  Tidy a cluttered directory — propose a folder structure, sort loose files into type folders (documents, images, audio, video, archives…), split photos from videos, flatten nested folders without name clashes, chunk a huge folder into numbered subfolders of N files, and consolidate overlapping folder

• files-permissions

  Diagnose and fix "permission denied" on Ubuntu — walk the path with namei, read ownership, modes and ACLs, check group membership (and the re-login trap after usermod), umask, setgid/sticky shared folders, mount options like noexec or CIFS uid mapping, and AppArmor or snap confinement denials — then

• hardware-gpu

  Inspect and fix GPU setup on Ubuntu — identify GPUs and the driver bound to each, install or switch NVIDIA drivers with ubuntu-drivers, check AMD amdgpu/ROCm and Intel stacks, Secure Boot/MOK signing for DKMS modules, Wayland vs X11 with NVIDIA, compute readiness for CUDA/ROCm workloads, GPU passthr

• hardware-power

  Configure sleep, hibernation, wake sources and Wake-on-LAN on Ubuntu — check suspend support, set up hibernation (swap size, resume device and offset, initramfs, Secure Boot caveat), stop spurious wake-ups, tune lid and idle behaviour in logind, choose power-profiles-daemon or TLP, and enable Wake-o

• hardware-profile

  Build a hardware profile of an Ubuntu machine that both people and agents can read — system identity, CPU, memory modules, storage, GPU, network, USB/PCI, firmware and sensors — saved as Markdown, with optional per-component deep dives, an optional hw-probe run, and a "will device X work with this m

• hardware-usb-input

  List and troubleshoot USB devices and input hardware on Ubuntu — the USB tree, identifying an unknown device, watching hot-plug events, autosuspend and power problems, reading keyboard/mouse scancodes with evtest or libinput, and remapping keys persistently with hwdb, keyd or input-remapper. Use whe

• host-workspace

  Create a working directory (optionally a private git repo) for administering one Ubuntu host or a small fleet over SSH. It holds a CLAUDE.md/AGENTS.md describing the host (SSH alias, role, desktop or server, key paths, services), context/ for profiles, backups/ for config copies and outputs/ for log

• network-cloudflared-tunnel

  Check and manage a Cloudflare Tunnel (cloudflared) on Ubuntu — install from Cloudflare's apt repository, service status and logs, locally-managed config.yml ingress rules or token-based remotely-managed tunnels, cloudflared tunnel list/info, validate ingress and test which rule a URL hits, check the

• network-config

  Change network configuration on Ubuntu safely — find the renderer (netplan + systemd-networkd on servers, NetworkManager on desktops), set a static IP, gateway, DNS and routes in /etc/netplan, apply with netplan try for automatic rollback, edit NetworkManager connections with nmcli, set the hostname

• network-diagnose

  Live network triage on Ubuntu — interfaces, link state and errors, addresses and DHCP, routes and default gateway, DNS through systemd-resolved, reachability of gateway, internet and named LAN hosts, MTU problems, Wi-Fi signal and association, listening services and port conflicts, and the NetworkMa

• network-lan-scan

  Discover the hosts on a local network from an Ubuntu machine — derive the subnet, sweep it with arp-scan or nmap ping scan, read the neighbour table, browse mDNS/Bonjour services with avahi, resolve names, look up vendors from MAC OUIs, and write a raw capture plus a human-readable Markdown LAN map.

• network-ssh-hosts

  Manage SSH host aliases and remote-admin access from an Ubuntu machine — list and add ~/.ssh/config entries, test reachability and non-interactive key auth (BatchMode), copy a public key to hosts with ssh-copy-id so the rest of these skills can run remotely, enable ControlMaster multiplexing for man

• packages-apt-health

  Diagnose and repair a broken APT/dpkg state on Ubuntu — held locks ("Could not get lock /var/lib/dpkg/lock-frontend"), interrupted installs, unmet dependencies, half-configured packages, held packages, NO_PUBKEY / signature errors, a full /var, and odd apt configuration. Use when asked "apt is broke

• packages-cli-tools

  Install and verify common admin and developer CLIs on Ubuntu by the preferred method for each — GitHub CLI (gh) from its official apt repo, AWS CLI v2, rclone, Backblaze b2, pipx, uv, yadm, SDKMAN and Homebrew on Linux (and when not to use it) — and produce a verification table showing installed, ve

• packages-install

  Install software on Ubuntu by the best available method — pick between the Ubuntu archive, a vendor apt repo, snap, flatpak, a .deb, an AppImage, a GitHub release binary or a source build, preferring whatever updates itself with least maintenance; install a downloaded .deb/AppImage/tarball; install

• packages-repos

  Audit and manage APT software sources on Ubuntu — list every repository (classic .list and deb822 .sources), identify PPAs and third-party vendor repos, find failing, duplicate or unsigned sources, migrate legacy apt-key keys to /etc/apt/keyrings with signed-by, add a vendor repo correctly, remove o

• packages-snap-flatpak

  Manage snaps and flatpaks on Ubuntu — list, refresh, hold or schedule snap refreshes, limit retained revisions and remove disabled ones to reclaim disk, inspect snap services and interface connections; set up Flathub, update, remove unused runtimes and adjust flatpak permissions with overrides. Use

• packages-updates

  Check for and apply Ubuntu updates and manage automatic updating — pending security vs regular updates, phased and kept-back packages, apt upgrade vs full-upgrade, reboot-required and needrestart, old-kernel cleanup, unattended-upgrades configuration (20auto-upgrades, 50unattended-upgrades, origins,

• performance-audit

  Repeatable, read-only performance audit of an Ubuntu machine, desktop or server — load vs cores, pressure stall (PSI), memory and active swapping, OOM kills, disk space and inodes, top processes, Docker containers near their memory limit or restart-looping, temperatures and failed units — ranked int

• services-check

  Sweep systemd service state on an Ubuntu machine — failed units (system and user), degraded state, flapping services with high restart counts, what is enabled at boot, leftover units pointing at removed software, and the last errors of each problem unit. Read-only. Use when asked "check services", "

• services-logs

  Fetch, save and summarise logs on an Ubuntu machine — journal by unit, priority and time window, a dated copy saved locally, repeating messages grouped, a "noteworthy, not just errors" review (restarts, OOM kills, kernel warnings, auth events), live following, classic /var/log files, Docker containe

• services-mail-relay

  Make an Ubuntu server able to send notification mail (from unattended-upgrades, cron, smartd, mdadm, backup scripts) through an external SMTP relay, using msmtp-mta (simple, no daemon) or Postfix as a satellite with relayhost and SMTP auth; then test sending, inspect the queue and read mail logs. Us

• services-manage

  Control and author systemd units on Ubuntu — start, stop, restart, reload, enable, disable, mask; override a packaged unit with a drop-in via systemctl edit; write a new service or a systemd timer (system or user, with lingering); set Restart= policies and resource limits (MemoryMax, CPUQuota); vali

• shell-aliases

  Manage bash aliases in ~/.bash_aliases on Ubuntu — list and search, add or edit with a conflict check, delete, prune duplicates and broken entries, suggest useful aliases, generate a Markdown reference, and take timestamped backups before every change. Use when asked "add an alias for X", "what alia

• shell-config

  Check and fix bash shell configuration on Ubuntu — validate ~/.bashrc, ~/.profile and ~/.bash_profile (syntax, shellcheck, slow startup), explain login vs non-login loading, analyse PATH (duplicates, missing directories, ~/.local/bin, ordering and shadowed binaries), set environment variables in the

• storage-backup

  Assess, audit and set up backups on an Ubuntu machine — decide what needs backing up, find existing backup jobs (restic, borg, rclone, Timeshift, Déjà Dup, cron or systemd timers), check when each last succeeded, list restore points, prove a restore works, and create a simple restic or rclone job on

• storage-drive-health

  Check the physical health of SATA, SAS, NVMe and USB-attached drives with SMART — overall verdict, the attributes that predict failure, self-tests, and continuous monitoring with smartd. Use when asked "is my disk dying", "check drive health", "SMART status", "NVMe wear", "run a disk self-test", "se

• storage-filesystems

  Check filesystem health on Ubuntu — mounts and fstab sanity, ext4/XFS/BTRFS integrity checks, inode exhaustion, kernel I/O errors, read-only remounts, and Snapper/Timeshift snapshot health. Use when asked "check the filesystem", "fsck", "is my fstab ok", "disk went read-only", "btrfs scrub", "xfs ch

• storage-network-mounts

  Set up, check and repair NFS and SMB/CIFS network mounts on Ubuntu — client packages, discovering exports and shares, testing a mount, persistent fstab or systemd automount entries that don't hang boot, credentials files, ownership mapping, and stale-handle recovery. Use when asked "mount the NAS",

• storage-raid-lvm

  Inspect and maintain software RAID (mdadm), LVM, ZFS pools and hardware RAID on Ubuntu — array state, degraded or rebuilding arrays, replacing a failed member, keeping mdadm.conf and the initramfs in sync, and growing a logical volume plus its filesystem. Use when asked "RAID status", "is my array d

• storage-reclaim-space

  Find and free disk space on a full or filling Ubuntu machine — quick hotspot scan, then the usual safe wins (journald vacuum, apt cache and orphaned packages, old kernels, disabled snap revisions, rotated logs, crash dumps, Docker) with a report before anything is deleted. Use when asked "disk is fu

• system-change-log

  Write a short, structured, dated record of an admin change or fix on an Ubuntu machine — problem, root cause, fix, how to reverse it — into a log directory, optionally committing it to git. It also writes longer host documentation (configuration, runbooks, filesystem layout) when asked. Use right af

• system-config-edit

  Safely change a system configuration file on an Ubuntu machine. It backs up the file first, makes the edit, validates it with the owning tool's checker (sshd -t, nginx -t, netplan generate, visudo -c, systemd-analyze verify, docker compose config and others), reloads only if validation passes, and c

• system-diagnose

  Triage a problem that is happening right now (or just happened) on an Ubuntu machine from fresh evidence — the last few minutes of the journal, kernel messages, failed units, OOM kills, coredumps — and return a diagnosis with the supporting log lines and a fix. Use when the user says "something just

• system-preflight

  Detect what kind of Ubuntu machine you are about to administer — Ubuntu version and LTS status, desktop or server, desktop environment and Wayland/X11, NetworkManager or systemd-networkd, VM or bare metal, passwordless sudo, and which tools (snap, flatpak, docker, smartctl, nvidia-smi) exist — and r

• system-resources

  Diagnose slowdowns, lag, high load, memory pressure, swap thrashing, runaway or zombie processes and I/O wait on an Ubuntu machine, and watch resources live. Use when the user says "the machine is slow", "laggy", "high load", "out of memory", "what is eating my RAM/CPU", "swap is full", "fans are lo

• system-status

  One-shot health snapshot of an Ubuntu machine — OS and kernel, uptime and load, memory and swap, disk and inode fill, failed units, recent errors, pending updates and reboot flag, plus Docker and desktop add-ons when present. Read-only. Use when asked "how is this box doing", "status of <host>", "he

• system-virtualization

  Check and set up virtualization on Ubuntu, in both directions. As a host, it covers CPU VT-x/AMD-V and BIOS enablement, KVM/QEMU/libvirt install, kvm/libvirt groups, the default network, storage pools, nested virtualization, IOMMU for PCI passthrough, and conflicts with VirtualBox or VMware. As a gu

Subagents (2)

• ubuntu-operator

  Autonomous operator for multi-step admin work on one or more Ubuntu machines, local or over SSH. Use it for jobs that chain many commands with checks in between, for example "update all my servers and reboot the ones that need it", "deploy this compose stack", "move this service's data to the new di

• ubuntu-troubleshooter

  Read-only investigator for a misbehaving Ubuntu machine, local or over SSH. It works through logs, services, resources, Docker, storage and network from evidence to a root cause, and returns a diagnosis with the supporting lines and a proposed fix without applying it. Use it when the cause is unknow

Similar plugins