HomeButler

Only the changes worth mentioning.
A single Go binary that remembers what your server looked like last time, and tells you — or an agent — what moved.

Website · Docs · Releases

brew install Higangssh/homebutler/homebutler     # or: curl -fsSL https://raw.githubusercontent.com/Higangssh/homebutler/main/install.sh | sh
homebutler report                                # first run saves a baseline; the second tells you what moved

Or as a container, reaching your machines over SSH — docs/docker.md: docker run -v ~/.config/homebutler:/config -v ~/.ssh:/root/.ssh:ro ghcr.io/higangssh/homebutler

Section rules, labels, and severities are colour-coded in a terminal. Colour is dropped automatically when output is piped, redirected, or run from cron.

That is the whole idea. Most homelab tools show you a graph of right now, and leave "does this matter?" to you. HomeButler remembers what your server looked like last time, decides what is worth saying, and says it — six containers before and six after is not "no change" when one of them is a different container.

Reading a change

Every line is three columns: what kind of change, what it happened to, and what exactly happened. The kind is one of eight words, and it is the same word in --json, so an agent branches on it without reading prose:

replaced is the one the rest of this exists for. A container recreated under the same name leaves every count identical, which is why a report that compares counts — as this one did before 0.26.0 — answers "no significant changes" while the thing you were running has been swapped out underneath you.

skipped is the second: homebutler says it could not compare rather than reporting nothing changed. An all-clear it cannot stand behind is worse than no answer.

The header names the snapshot being compared against, so "what changed" is never ambiguous about the window it covers.

📖 What earns a line, and what is deliberately suppressed →

HomeButler helps you answer the boring but painful questions every homelab eventually creates:

• What is running on my server right now?
• Which container owns this port?
• Why did this service restart at 3 AM?
• Is my backup actually restorable?
• Can I install this self-hosted app without hand-writing another compose file?
• Can I let an AI assistant inspect my server without handing it a full SSH shell?

No daemon required, no database. Run it from a terminal, cron, a web dashboard you start when you want one, or an AI agent.

The design goal is simple: give humans and agents a narrow, structured interface to the server. HomeButler returns readable summaries and JSON instead of asking you to trust a black-box shell session.

a run through homebutler: report naming what moved since the last snapshot — a container gone, a port answered by something new, one recreated behind the same name — then the same answer as JSON with the kind as a field, then doctor listing what is exposed and what has never been backed up with the command for each, then a backup drill booting the archive in its own container and requiring an HTTP health check, then the same report on a phone, then the settings screen editing notification channels

The whole thing in half a minute · watch it full size

Quick Start

# One-line install (auto-detects OS/arch)
curl -fsSL https://raw.githubusercontent.com/Higangssh/homebutler/main/install.sh | sh

# Or via Homebrew
brew install Higangssh/homebutler/homebutler

# Or as a container — the hub that reaches your machines over SSH
docker run -d -p 8080:8080 -v ~/.config/homebutler:/config -v ~/.ssh:/root/.ssh:ro \
  ghcr.io/higangssh/homebutler serve --host 0.0.0.0 --token "$(openssl rand -hex 16)"

# Interactive setup — add your servers in seconds
homebutler init

Use it right away:

homebutler status                    # CPU, memory, disk, uptime
homebutler docker list               # running containers
homebutler inventory scan            # containers + ports + topology
homebutler report                    # butler-style health report + change summary
homebutler install uptime-kuma       # deploy a self-hosted app
homebutler backup drill uptime-kuma  # verify a backup actually restores
homebutler watch tui                 # terminal dashboard
homebutler serve                     # web dashboard at http://localhost:8080

Machine-readable output is available everywhere:

homebutler status --json
homebutler inventory scan --json
homebutler report --json

What it does

• Install apps — deploy Uptime Kuma, Jellyfin, Pi-hole, Gitea, Portainer, and more with one command
• Map your server — see containers, exposed ports, system ports, and service topology
• Run a doctor check — diagnose resource pressure, stopped containers, public ports, backup hygiene, notifications, report baseline readiness, and configured Proxmox endpoint reachability
• Catch crashes — save logs before/after Docker, systemd, or PM2 restarts and detect flapping loops
• Verify backups — boot backups in isolated containers before you trust them
• See a Proxmox cluster — nodes, QEMU and LXC guests, storage, task status, and honest dashboard freshness, with power actions that name their target explicitly
• Use it anywhere — CLI, JSON, web dashboard, or MCP for AI agents without giving them SSH

Why homebutler?

Self-hosting is not hard because one docker compose up is hard. It is hard because the maintenance never ends: ports collide, containers restart silently, backups look fine until restore day, and every server becomes a slightly different snowflake.

HomeButler is a small operations toolkit for that messy middle.

Alongside what you already run

Keep Uptime Kuma for is it up, Beszel or Netdata for the graph, Dozzle for the logs. HomeButler answers the question none of them ask: what is different from last time, and does it matter?

• No agent on the machines it watches. The binary is put there once with homebutler deploy and runs only when asked, over SSH — no daemon, no open port, nothing listening between runs.

• The judgement is a written rule, not a model. What earns a line is in docs/report.md, and the same input gives the same report — no AI required, no account, no paid tier.

• It checks that a backup comes back. A backup you have never restored is a folder. backup drill boots the archive as a second copy of the app, on a network and port of its own, and requires it to answer an HTTP health check before calling the archive good — then removes everything it made. A failed drill exits non-zero, so backup drill --all belongs in the same cron entry as backup. docs/backup.md has what it does and what it does not prove.

  code

  🔐 Integrity: ✅ tar valid (8 files)
  🚀 Boot: ✅ container started in 0s
  🌐 Health: ✅ HTTP 200 on port 60405
  
  ✅ DRILL PASSED
  

Core workflows

🧾 Butler Report

homebutler report
homebutler report --keep 7      # retain only the latest 7 snapshots
homebutler report --no-save     # preview without writing a snapshot

report gives you a concise butler-style summary of your homelab: current health, warnings, notable changes since the previous snapshot, and suggested next commands. On the first run, HomeButler creates a baseline under ~/.homebutler/reports/snapshots/; later runs compare against the latest snapshot. Old snapshots are pruned automatically (--keep 30 by default) so reports do not grow forever.

🩺 Doctor Check

homebutler doctor
homebutler doctor --strict          # non-zero exit if warnings/failures are found
homebutler doctor --json            # automation / MCP friendly

doctor is a read-only preflight for the problems homelab users usually discover too late: high disk or memory usage, stopped containers, public bind ports, stale or missing backups, missing notifications, whether report has a baseline for change detection, and whether each configured Proxmox endpoint is reachable with the token it has. Every finding names the next command to run, so --strict makes it usable from cron or CI — including a Proxmox host that is unreachable or rebooting.

Every finding carries a category, and --json carries the same word, so a caller filters on it without reading the title:

incomplete is the one worth handling. It means the answer you are reading is missing a section, which is different from that section being fine.

🗂 Config Validation

homebutler config validate
homebutler config validate --strict   # exit non-zero on warnings too
homebutler config validate --json

config validate reads your config without starting anything and tells you which file was used, which of the four resolution rules picked it, and what homebutler actually made of each section. It exists because the two ways config goes wrong are both silent: a key homebutler does not recognise is dropped without a word, and a --config path that does not exist falls back to built-in defaults rather than failing.

Sections
   ✓ servers     2 servers (homelab, nas)
   · notify      not set
   ✓ alerts      cpu 95% · memory 85% · disk 90%

Findings
   ⚠️ Line 5: field notifiy not found in the homebutler config
      → Did you mean "notify"? Unrecognised keys are ignored silently.

📦 One-Command App Install

$ homebutler install list
📦 Available apps (15):

  uptime-kuma          A self-hosted monitoring tool
  vaultwarden          Lightweight Bitwarden-compatible password manager
  jellyfin             Free software media system for streaming movies, TV, and music
  pi-hole              Network-wide ad blocking via DNS filtering
  …

homebutler install uptime-kuma — Deploy self-hosted apps in seconds. Pre-checks Docker, ports, and duplicates. Generates docker-compose.yml automatically. See all available apps →

🗺️ Inventory & Topology

homebutler inventory scan
homebutler inventory show --filter exposed
homebutler inventory export --format mermaid
homebutler --json inventory scan

inventory scan gives you a quick map of what is running on a server: system health, Docker containers, app ports, and system ports. Docker-published ports are connected back to the container that owns them, so local forwarding details like Colima/Lima stay understandable.

🏠 Home Network
   Server  homelab (192.168.1.10)
   Summary ✅ 1 running · ⚪ 1 stopped · 🌍 2 public ports · 🔒 4 local ports

📦 Containers (2)
   ├─ ⚪ vaultwarden · not started
   │  └─ image vaultwarden/server:latest
   └─ ✅ api-server · running
      ├─ image my-api:latest
      └─ exposes :8080 → 8080/tcp

🌐 App Ports (1)
   └─ 🌍 :8080/tcp · api-server

To answer "what is reachable from outside my machine/network?" without reading the whole tree, filter the scan to exposed ports only:

homebutler inventory scan --filter exposed

🏠 Home Network
   Server  homelab

🌐 Exposed Ports
   ├─ :8080/tcp · api-server
   └─ :8443/tcp · dashboard

Only ports listening on all interfaces (0.0.0.0, ::, *) are shown. Anything bound to a specific address is hidden, including loopback and LAN addresses. Unsupported filter values return an error, as does combining --filter with --json; the default inventory scan output is unchanged.

Use Mermaid export when you want a diagram for GitHub, Obsidian, docs, or an AI assistant:

graph TD
  home["🏠 Home Network"] --> homelab["🖥 homelab<br/>192.168.1.10"]
  homelab --> c1["📦 api-server<br/>running"]
  homelab --> p1["🌍 :8080/tcp<br/>api-server"]
  c1 -. exposes .-> p1

Demo

🌐 Web Dashboard

homebutler serve — A web dashboard embedded in the single binary via go:embed. Servers, Docker containers, open ports, alerts and Wake-on-LAN from any browser. With --token it also edits the config: thresholds, notification channels, servers, Proxmox endpoints and wake devices.

The Report tab, where you actually read it

the Report tab on a phone: what needs attention first — a container recreated and not running, a port answered by something new — then the changes with their kind word, including a container replaced behind the same name and a comparison that could not be made, then doctor's findings ordered by severity with the command to run for each

The screen somebody opens after a notification. What needs attention first, then what moved with the kind word each change carries, then the snapshot being compared against and how old it is. A comparison that could not be made says so rather than reading as an all-clear.

Loading it saves nothing — a page that took a snapshot every time it was opened would prune the baseline you wanted to compare against. Saving one is a button.

homebutler serve              # Start on port 8080
homebutler serve --port 3000  # Custom port
homebutler serve --demo       # Demo mode with realistic sample data

🔄 Process Restart Watch

Your container crashed at 3 AM — but why? homebutler watch catches it the moment it happens, saves the dying logs, figures out the cause, and tells you if it's happening over and over.

Supported backends: Docker (real-time event stream) · systemd (polling) · PM2 (polling)

Step 1: Add targets to watch

homebutler watch add nginx              # Interactive: choose Docker / systemd / PM2
homebutler watch add --kind docker nginx          # or specify directly
homebutler watch add --kind systemd nginx.service
homebutler watch add --kind pm2 my-api
homebutler watch list                   # See what you're watching

Step 2: Start monitoring

homebutler watch start                  # Foreground, Ctrl+C to stop
homebutler watch start --interval 10s   # Custom poll interval (default 30s)

homebutler watch install     # register it with systemd or launchd
homebutler watch installed   # is it registered?
homebutler watch uninstall

watch install hands the loop to whatever supervises the host — a systemd user unit on Linux, a launchd agent on macOS — so monitoring survives logout and reboot. Both are user-level and neither is a preference: on Linux the watch list lives in your home directory, so a root unit would find an empty list; on macOS Docker Desktop only runs inside a logged-in session, so a LaunchDaemon would poll a daemon that is not there. On Linux a user unit stops at logout unless you run sudo loginctl enable-linger $USER, which watch install tells you.

watch start is the monitoring process. It watches the containers and services on the watch list for restarts, checks CPU, memory and disk against your thresholds, and runs any remediation rules you have configured — one process, one set of notification providers. alerts --watch still exists and does the threshold half on its own.

Every endpoint under proxmox: in your config is polled too: unreachable or ACL-filtered endpoints and any guest listed under that endpoint's guests: report one incident when the problem starts and one recovery incident when it clears. A guest not listed there is observational only — watch start never alerts on it, deliberately stopped or not. See Proxmox setup → for the guests: field.

When a crash is detected, you'll see:

[03:14:22] INCIDENT: nginx (incident nginx-20260410-031422.581-7a2124)
  Crash: OOM — process killed by SIGKILL (oom, confidence: high)
  ⚠ FLAPPING: acute (3 restarts in short window)

Step 3: Investigate

homebutler watch history                # List all incidents
homebutler watch show <incident-id>     # Full details

watch show output includes:

• Pre-death logs — what the process printed right before it died
• Post-restart logs — what happened after the restart
• Crash analysis — category (oom / panic / segfault / timeout / dependency / error), reason, confidence level, matched log patterns
• Flapping status — if the process is stuck in a crash loop

Crash Analysis

Every incident is automatically analyzed using exit codes and log patterns:

Log patterns like panic:, Out of memory, Connection refused, FATAL, and timeout are matched automatically to help identify the root cause.

Flapping Detection

Detects when a process is stuck in a restart loop (e.g., crash → restart → crash again):

• Acute — 3+ restarts within 10 minutes (something is broken right now)
• Chronic — 5+ restarts within 24 hours (slow recurring issue)

Flapping incidents are tagged [FLAPPING] in history and highlighted in watch show.

Notifications (optional, off by default)

Notifications are disabled by default, which is useful for air-gapped or closed networks where everything runs locally.

A minimal example in ~/.config/homebutler/config.yaml:

notify:
  telegram:
    bot_token: "your-bot-token"
    chat_id: "your-chat-id"

watch:
  enabled: true
  notify_on: flapping
  cooldown: 5m
  flapping:
    short_window: 10m
    short_threshold: 3
    long_window: 24h
    long_threshold: 5
  retention:
    max_incidents: 200

alerts:
  cpu: 90
  memory: 85
  disk: 90
  rules:
    - name: cpu-spike
      metric: cpu
      threshold: 90
      action: notify

    - name: elsa-monitor-down
      metric: container
      kind: systemd          # docker (default) | systemd | pm2
      watch: [lh-elsa-monitor.service]
      action: restart

Restarting things that are not containers

action: restart restarts Docker containers unless the rule says otherwise. kind: systemd or kind: pm2 points it at a service or a PM2 app instead.

The kind is written on the rule rather than looked up from the watch list, so restarting a host service is something you asked for in the config. It also means every rule written before kind existed keeps meaning exactly what it meant.

Two things worth knowing before using it:

systemctl restart needs root or a polkit rule. Running homebutler unprivileged, a systemd restart will be refused, reported as failed, and warned about when alerts --watch starts rather than when the rule first fires.

A target that is flapping is not restarted. Restarting something already in a restart loop feeds the loop, and most systemd units carry Restart=always, so homebutler restarting them fights systemd's own backoff. The thresholds are the watch.flapping ones above, and the skip is reported rather than counted as either success or failure. This applies to Docker targets too.

Legacy ~/.homebutler/watch/config.json is still read as a fallback for watch-specific settings, and legacy alerts.yaml notify/webhook provider settings are still accepted for older setups.

• watch.enabled: true — allow watch notifications

• watch.notify_on: flapping — notify only when repeated restart loops are detected

• watch.notify_on: incident — notify on every incident

• watch.notify_on: all — notify on both incidents and flapping

• watch.notify_on: off — disable watch notifications without removing provider config

• watch.cooldown: 5m — suppress duplicate notifications for the same event fingerprint during the cooldown window

• watch.flapping — optional advanced tuning for restart-loop detection

• watch.retention.max_incidents: 200 — how many incidents to keep on disk, newest first. The directory grows fastest exactly when a service is restarting in a loop. Set -1 to keep everything.

  Each incident keeps up to 100 captured log lines per side, and at most 64 KB of them. Line counts alone do not bound a file: one stack trace or JSON document on a single line is arbitrarily long, and a container being OOM-killed is exactly the one likely to write one. A log that does not fit keeps its end — the last thing a process said is what explains why it stopped — and says how much was dropped.

These settings can also be written under a watch.notify: block, which is the canonical form:

watch:
  notify:
    enabled: true
    notify_on: flapping
    cooldown: 5m
  flapping:
    short_window: 10m

Both spellings are read, so either layout works. If a file contains both, the notify: block wins and homebutler config validate says so.

Manage targets

homebutler watch remove nginx           # Stop watching
homebutler watch check                  # One-shot check (no continuous monitoring)

🧊 Proxmox VE

homebutler proxmox status
homebutler proxmox guests --status running
homebutler proxmox guest shutdown --node pve1 --type lxc --vmid 105 --confirm
homebutler proxmox task UPID:pve1:... --node pve1

A Proxmox endpoint is its own kind of target, configured under proxmox: with an API token rather than SSH, so it does not join the --server or --all fan-out. TLS verification stays on: trust comes from a pinned SHA-256 fingerprint, then a CA file, and only then an explicit insecure fallback.

Reads are plain. Power actions are not: every one of them takes an explicit endpoint, node, guest type and VMID, and refuses to run without --confirm, which is checked before any credential is read. They also need their own action_token_id (plus action_token or action_token_file) configured on the endpoint — the read token alone will not start, reboot, or shut down a guest; see Proxmox setup → for creating that second token. shutdown asks the guest to shut down cleanly — it is not Proxmox's hard stop, which cuts power and can leave a filesystem behind it. A successful action reports the task it submitted, not that the guest finished; proxmox task answers that separately.

proxmox script prints the install command for a Community Script pinned to one commit, along with a warning that the script is not reviewed by homebutler and runs as root. It never fetches or runs it — see #62 for why that line is where it is.

📖 Proxmox setup, tokens, and TLS →

🖥️ TUI Dashboard

homebutler TUI dashboard

homebutler watch tui — A terminal-based dashboard powered by Bubble Tea. Monitors all configured servers with real-time updates, color-coded resource bars, and Docker container status. No browser needed.

🧠 AI-Powered Management (MCP)

Use natural language when you want automation. MCP clients can call homebutler tools to check server status, list Docker containers, inspect ports, or run operational workflows. See screenshots & setup →

App Install

Deploy self-hosted apps with a single command. Each app runs via docker compose with automatic pre-checks, health verification, and clean lifecycle management.

# List available apps
homebutler install list

# Install (default port)
homebutler install uptime-kuma

# Install with custom port
homebutler install uptime-kuma --port 8080

# Install jellyfin with media directory
homebutler install jellyfin --media /mnt/movies

# Check status
homebutler install status uptime-kuma

# Stop (data preserved)
homebutler install uninstall uptime-kuma

# Stop + delete everything
homebutler install purge uptime-kuma

How it works

~/.homebutler/apps/
  └── uptime-kuma/
       ├── docker-compose.yml   ← auto-generated, editable
       └── data/                ← persistent data (bind mount)

• Pre-checks — Verifies docker is installed/running, port is available, no duplicate containers
• Compose-based — Each app gets its own docker-compose.yml you can inspect and customize
• Data safety — uninstall stops containers but keeps your data; purge removes everything
• Cross-platform — Auto-detects docker socket (default, colima, podman)

Available apps

App-specific options

# Jellyfin: mount your media library
homebutler install jellyfin --media /mnt/movies

# Pi-hole / AdGuard: DNS ad blocking (port 53 required)
homebutler install pi-hole
# ⚠️ If port 53 is in use (Linux): sudo systemctl disable --now systemd-resolved

# Portainer: Docker GUI (mounts docker socket)
homebutler install portainer
# Access via HTTPS: https://localhost:9443

# Nginx Proxy Manager: reverse proxy
homebutler install nginx-proxy-manager
# Default login: admin@example.com / changeme (change immediately!)

# Any app: custom port
homebutler install <app> --port 9999

Safety checks

• Port conflict detection — Checks if the port is already in use before install
• DNS mutual exclusion — Warns if pi-hole and adguard-home are both installed
• Docker socket warning — Alerts when an app requires Docker socket access (portainer)
• OS-specific guidance — Linux gets systemd-resolved fix, macOS gets lsof command
• Post-install tips — DNS setup, HTTPS access, default credential warnings

Want more apps? Open an issue or see Contributing.

Usage

homebutler <command> [flags]

Commands:
  status              System status (CPU, memory, disk, uptime)
  doctor              Diagnose health, exposure, backups, and readiness
  config validate     Check the config file and report what is ignored
  docker list         List running containers
  install <app>       Install a self-hosted app (docker compose)
  alerts              Show current alert status
  watch tui           TUI dashboard (monitors all configured servers)
  watch add/list/remove  Manage watched containers
  watch check/start   One-shot or continuous restart detection
  watch history/show  Browse restart history
  proxmox status      Proxmox VE cluster, nodes, guests, and storage
  serve               Web dashboard (browser-based, go:embed)

Flags:
  --json              JSON output (default: human-readable)
  --verbose, -v       Show detailed error information
  --server <name>     Run on a specific remote server
  --all               Run on all configured servers in parallel
  --port <number>     Port for serve command (default: 8080)
  --config <path>     Config file (auto-detected, see Configuration)

Run homebutler --help for all commands.

Commands:
  init                Interactive setup wizard
  config validate     Check the config file and report what is ignored
  status              System status (CPU, memory, disk, uptime)
  doctor              Diagnose health, exposure, backups, and readiness
  watch tui           TUI dashboard (monitors all configured servers)
  watch add <name>    Add container to restart watch list
  watch list          Show watched containers
  watch remove <name> Remove container from watch list
  watch check         One-shot restart check
  watch start         Continuous monitoring: restarts, thresholds, rules
  watch install       Register watch with systemd or launchd
  watch installed     Report whether it is registered
  watch uninstall     Remove the service unit
  watch history       List restart history (alias: incidents)
  watch show <id>     Show restart details with logs
  serve               Web dashboard (browser-based, go:embed)
  docker list         List running containers
  docker restart <n>  Restart a container
  docker stop <n>     Stop a container
  docker logs <n>     Show container logs
  docker top <n>      Show processes running inside a container
  docker inspect <n>  Show image, state, ports, mounts, networks, health
  report              What changed since the last snapshot
  inventory scan      Map containers, ports, and topology
  inventory show      Same as scan (--filter exposed narrows it)
  inventory export    Export the map (--format mermaid)
  proxmox status      Proxmox VE cluster, nodes, guests, storage
  proxmox guests      List QEMU and LXC guests
  proxmox node <n>    Node detail
  proxmox guest ...   start / shutdown / reboot (needs --confirm)
  proxmox task <upid> Task status for an action already submitted
  proxmox tasks       Recent tasks on a node
  proxmox script      Community Script install commands (prints, never runs)
  notify test         Send a test notification through configured providers
  wake <name>         Send Wake-on-LAN packet
  ports               List open ports with process info
  ps                  Show top processes (alias: processes)
  ps --sort mem       Sort by memory instead of CPU
  ps --limit 20       Show top 20 (default: 10, 0 = all)
  network scan        Discover devices on LAN
  alerts              Show current alert status
  alerts --watch      Thresholds only (watch start covers these too)
  trust <server>      Register SSH host key (TOFU)
  backup              Backup Docker volumes, compose files, and env
  backup list         List existing backups
  backup drill <app>  Verify backup restores correctly (isolated)
  backup drill --all  Verify all apps in backup
  restore <archive>   Restore from a backup archive
  upgrade             Upgrade local + all remote servers to latest
  deploy              Install homebutler on remote servers
  install <app>       Install a self-hosted app (docker compose)
  install list        List available apps
  install status <a>  Check installed app status
  install uninstall   Stop app (keep data)
  install purge       Stop app + delete all data
  mcp                 Start MCP server (JSON-RPC over stdio)
  version             Print version

Flags:
  --json              JSON output (default: human-readable)
  --verbose, -v       Show detailed error information
  --server <name>     Run on a specific remote server
  --all               Run on all configured servers in parallel
  --port <number>     Port for serve command (default: 8080)
  --demo              Run serve with realistic demo data
  --watch             Continuous monitoring mode (alerts command)
  --interval <dur>    Watch interval, e.g. 30s, 1m (default: 30s)
  --config <path>     Config file (auto-detected, see Configuration)
  --local             Upgrade only the local binary (skip remote servers)
  --local <path>      Use local binary for deploy (air-gapped)
  --service <name>    Target a specific Docker service (backup/restore)
  --allow-bind <path> Host path a restore may write a bind mount to (repeatable)
  --endpoint <name>   Proxmox endpoint from config (optional if only one)
  --confirm           Required for a Proxmox guest power action
  --to <path>         Custom backup destination directory
  --archive <path>    Specific backup archive for drill
  --all               Verify all supported apps (backup drill)

homebutler serve                # http://localhost:8080
homebutler serve --port 3000    # custom port
homebutler serve --demo         # demo mode with sample data

Backup & Restore

One-command Docker backup — volumes, compose files, and env variables.

homebutler backup                          # backup everything
homebutler backup --service jellyfin       # specific service
homebutler backup --to /mnt/nas/backups/   # custom destination
homebutler backup list                     # list backups
homebutler restore ./backup.tar.gz         # restore

⚠️ Database services should be paused before backup for data consistency.

📖 Full backup documentation → — how it works, archive structure, security notes.

Alert Thresholds (Advanced)

alerts still exists for CPU, memory, and disk threshold checks, but it is an advanced flow and not the recommended first step for new users.

homebutler alerts --watch                  # default: 30s interval
homebutler alerts --watch --interval 10s   # check every 10 seconds
homebutler alerts history                  # view alert history
homebutler notify test                     # test your notification channels

Default thresholds: CPU 90%, Memory 85%, Disk 90%. Start with watch, then add alerts only if you specifically want threshold-based checks.

🔍 Backup Drill

"Having a backup" and "being able to restore" are different things.

Backup Drill boots your backup in an isolated Docker environment and verifies the app actually responds — like a fire drill for your data. A failed drill exits non-zero.

homebutler backup drill uptime-kuma        # verify one app
homebutler backup drill --all              # verify all apps
homebutler backup drill --json             # machine-readable output
homebutler backup drill --archive ./file   # use a specific backup

Full walkthrough, including what a passing drill does not prove: docs/backup.md.

What happens:

1. Finds the latest backup archive
2. Verifies archive integrity (tar validation)
3. Creates an isolated Docker network + random port
4. Boots the app from backup data
5. Runs an HTTP health check
6. Reports pass/fail and cleans up everything

🔍 Backup Drill — uptime-kuma

  📦 Backup: ~/.homebutler/backups/backup_2026-04-04_1711.tar.gz
  📏 Size: 18.6 MB
  🔐 Integrity: ✅ tar valid (8 files)

  🚀 Boot: ✅ container started in 0s
  🌐 Health: ✅ HTTP 200 on port 58574
  ⏱️  Total: 2s

  ✅ DRILL PASSED

Zero risk — runs in a completely isolated environment. Your running services are never touched.

Supports health checks for: nginx-proxy-manager, vaultwarden, uptime-kuma, pi-hole, gitea, jellyfin, plex, portainer, homepage, adguard-home.

Configuration

homebutler init    # interactive setup wizard

📖 What report compares → — what earns a line, what is deliberately suppressed, and why.

📖 Configuration details → — config file locations, watch/notify options, and advanced alert thresholds.

Multi-server

Manage multiple servers from a single machine over SSH.

homebutler status --server rpi     # query specific server
homebutler status --all            # query all in parallel
homebutler deploy --server rpi     # install on remote server
homebutler upgrade                 # upgrade all servers

📖 Multi-server setup → — SSH auth, config examples, deploy & upgrade.

MCP Server

Built-in MCP server — an agent gets the same report as typed JSON, kind, target and detail rather than a sentence to parse, and every tool is classed read, write or destructive.

{
  "mcpServers": {
    "homebutler": {
      "command": "npx",
      "args": ["-y", "homebutler@latest"]
    }
  }
}

Works with Claude Desktop, ChatGPT, Cursor, Windsurf, and any MCP client.

📖 MCP server setup → — supported clients, available tools, agent skills.

Installation

Homebrew (Recommended)

brew install Higangssh/homebutler/homebutler

Automatically installs to PATH. Works on macOS and Linux.

One-line Install

curl -fsSL https://raw.githubusercontent.com/Higangssh/homebutler/main/install.sh | sh

Auto-detects OS/architecture, downloads the latest release, and installs to PATH.

npm (MCP server)

npm install -g homebutler

Downloads the Go binary automatically. This package publishes one command, homebutler-mcp, and it starts the MCP server — npx -y homebutler@latest launches that, not the CLI. For the command line, use one of the other installs above.

Go Install

go install github.com/Higangssh/homebutler@latest

Build from Source

git clone https://github.com/Higangssh/homebutler.git
cd homebutler
make build-all

make build-all compiles the web dashboard into the binary and needs Node installed. make build skips it — the CLI is complete either way, and homebutler serve then says the dashboard is missing and names the two ways to get it.

Uninstall

rm $(which homebutler)           # Remove binary
rm -rf ~/.config/homebutler      # Remove config (optional)

Architecture

The butler watches, decides what is worth saying, and says it — to you or to an agent.

Something changes → report names it and ranks it → a notification, the dashboard, or an MCP call carries the same answer.

        ┌─────────┐  ┌─────────┐  ┌─────────┐
        │   CLI   │  │   MCP   │  │   Web   │
        │ stdout  │  │  stdio  │  │  :8080  │
        └────┬────┘  └────┬────┘  └────┬────┘
             └────────────┼────────────┘
                          ▼
                   internal/*
         system · docker · ports · network
         wake · alerts · remote (SSH)

Three interfaces, one core:

All three call the same internal/ packages — no code duplication.

An agent reads the same report a person does, and gets the parts it has to branch on rather than a screen it has to interpret:

report = json.loads(subprocess.run(
    ["homebutler", "report", "--json"], capture_output=True, text=True).stdout)

for change in report["notable_changes"]:
    if change["kind"] == "replaced":         # the container came back as something else
        redeploy(change["target"])
    else:
        log(change["text"])                  # "port: :8080/tcp — nginx → caddy"

Every line carries text as well, so a caller that only wants to print it does not have to put the sentence back together.

Nothing above this is homebutler's business: an MCP client, a chat bot, a cron line or a person at a terminal all reach the same answer.

No ports opened by default. CLI and MCP use stdin/stdout only. The web dashboard is opt-in (homebutler serve, binds 127.0.0.1).

Contributing

Contributions welcome! Please open an issue first to discuss what you'd like to change. CONTRIBUTING.md covers what homebutler accepts and what a new target has to prove.

Security

Found a vulnerability? Report it privately through the Security tab rather than a public issue. SECURITY.md covers what is in scope and what to expect.

License

MIT