Changelog

What's new in SpinupTUI

Every release, newest first. The full technical record is in CHANGELOG.md.

v1.3.0

Modules: Beacon, Scheduler and Sentinel

November 10, 2026

SpinupTUI now keeps working after you close it.

Modules are parts of SpinupTUI that run on their own, on a Mac that stays awake. We call that Mac a station. Three ship today, and each builds on the one before:

  • Sentinel watches your fleet and works out what broke.
  • Beacon pages you.
  • Scheduler runs follow-ups on a clock.

Set them up from the new Modules tab (6). It shows what each step will change before it changes anything.

πŸ“£ The story behind this release: SpinupTUI 1.3.0: modules that keep watch after you close it

β—ˆ Sentinel: know why, not just that it's down

Every 5 minutes, Sentinel checks Uptime Kuma and groups whatever went down by server. Then it pages you and writes it up.

  • Bring an agent CLI (Claude Code or Codex) for a full diagnosis. Without one, you still get the evidence.
  • The agent can only read. It reaches servers through ssh-exec, and it can't post anything.
  • Sentinel files the report on your team's Basecamp board and pages you a one-line summary that names the site to blame.
  • A repeat within 6 hours becomes a follow-up on the same report, not a new one.

And a morning digest that only speaks up about what's new. At 08:00 it covers station health, the last 24 hours, server updates, and things monitoring can't see, like a crawler that stopped arriving. When nothing needs you, it stays quiet.

β—‰ Beacon: one command that pages you

  • spinuptui beacon send works from any script, cron job or agent. It pages you on Telegram, with retries and a send log.
  • Guided setup makes the Telegram bot with you.
  • Your pager answers back. Send /jobs to see the station's jobs, extend one that's about to expire, or run a failed one again.

β—· Scheduler: follow-ups that run themselves

  • Queue "check this tomorrow at 8" from any session, and the result arrives through Beacon.
  • A job can be a SpinupTUI command, a script, or a read-only agent prompt.
  • Every job expires, so a forgotten one stops by itself.
  • Station heartbeat: each tick pings an Uptime Kuma monitor you run elsewhere, so you notice when the station itself stops.

Also new

  • Choose how often Kuma checks a site: 1, 2, 5 or 10 minutes, from the monitoring overlay (M). Thanks to @lumpysimon for the request.
  • Headless commands start about 3Γ— faster. That matters for something started every five minutes.

Fixed

  • A reboot's Kuma maintenance window now ends by itself after 30 minutes. Before, quitting mid-reboot could leave monitors in Maintenance, where they never alert.

Notes

  • Stations are macOS-only for now. The rest of SpinupTUI is unchanged.
  • Basecamp reports post as whoever the Basecamp CLI is logged in as. We recommend a separate Basecamp user. How to set that up.
  • On a 1.3.0 alpha? There's nothing to migrate. Move from @next to @latest, or to Homebrew.

Docs: Sentinel Β· Scheduler Β· Beacon Β· spinuptui.com

View on GitHub β†’

v1.2.2

Kuma behind proxies, www-canonical pulls, mixed-stack media

November 6, 2026

A small fix release: three bugs, all found on real sites. Uptime Kuma connects through proxies that block WebSockets, pull db stops leaving production URLs behind, and the media fallback works while a site is mid-conversion to Bedrock.

Fixed

  • Uptime Kuma connects through proxies that block WebSockets. SpinupTUI was meant to fall back to plain HTTP when a WebSocket failed, but the fallback never actually ran. So a reverse proxy or CDN that doesn't pass WebSocket upgrades failed the whole connection, though Kuma also answers over HTTP. It now retries over HTTP. If both attempts fail, the error gives each one's real reason (e.g. Expected 101 status code (WebSocket); HTTP 404 (polling)), where before it said only websocket error.
  • pull db rewrites a www. site's URLs. The rewrite started from the SpinupWP domain. So a site whose domain is example.com but whose WordPress home is https://www.example.com kept every www. URL pointing at production, even though every step showed βœ“. The pull now asks production for its home URL over the same SSH connection as the export, and rewrites that as well as the domain. Post-import hooks get the home URL's host as SYNC_REMOTE_HOST, and every rewritten host in the new SYNC_REMOTE_HOSTS.
  • pull db rewrites URLs in plugin tables too. wp-cli's search-replace covers only WordPress's core tables unless told otherwise. So Yoast's indexables (and with them the local copy's canonical and schema URLs), cache and log tables, and form entries all kept production URLs. The rewrite now covers every table with the site's prefix.
  • The production media fallback (m) works when production runs a different stack. When a Standard WP site had been converted to Bedrock locally but production hadn't been switched yet, every missing image pointed at production's /app/uploads, which doesn't exist there, while the dialog said ON. The plugin (now v3) maps the local uploads path to production's, based on production's stack. The m dialog shows the mapping when the two differ. Press u in the m dialog to update an installed plugin.

View on GitHub β†’

v1.2.1

Fresh WordPress versions, and w opens the right page

October 27, 2026

A small fix release. WordPress versions stay current without you doing anything, and w opens the site or server you picked in SpinupWP instead of the dashboard.

Fixed

  • WordPress versions no longer go stale. SpinupTUI read a site's WordPress version once, when it first identified the site, and never again, so sites updated outside SpinupTUI (auto-updates, wp-admin, deploys) kept showing an old version, and r didn't help. Versions older than a day are now re-checked in the background whenever the site list loads. Each server's sites are checked one at a time, and the first failure on a server stops it for the day, so unattended SSH can't trip fail2ban. A reading that couldn't be refreshed shows its age in the Stacks list (WordPress 6.9.4 4mo), and the Details pane always says when it was taken.
  • w opens the right page. Links into SpinupWP need your account slug, which SpinupWP's API doesn't provide and setup never asked for, so unless you'd added accountSlug to your config by hand, w fell back to the dashboard. Now w asks for the slug the first time it needs it (pasting any dashboard URL works), saves it and opens the page you asked for. First-run setup also asks for it, as an optional step.

View on GitHub β†’

v1.2.0

Take sites offline until you can patch

October 21, 2026

When a server can't be patched yet against a critical vulnerability, the advice is to take its sites offline until it can be. SpinupTUI now does that a whole server at a time, or one site at a time, and remembers exactly what it took down, so bringing everything back is one step and never touches a site you'd disabled for some other reason.

Highlights

  • Take a server's sites offline (a). The server actions menu has a new last row, Take all N sites offline. It uses SpinupWP's own disable setting: nginx stops answering for every domain on those sites, so visitors can't connect and no request reaches PHP, including requests aimed straight at the server's IP. Files, databases and certificates are untouched. On a server SpinupTUI took offline, the same row becomes Bring N sites back online.
  • Or one site (O). Press O on a site in Servers, Search or Stacks, and O again to bring it back. PHP is shared by every site on a server, so the confirmation tells you how many other sites are still up and points you to a when it's the server you need to protect.
  • It brings back exactly what it took down. Each site is recorded the moment SpinupWP accepts its disable, so an interrupted run can still be undone. A site disabled some other way stays disabled, a failed re-enable stays recorded so running it again finishes the job, and sites taken offline one at a time come back with their server.
  • From the command line too. spinuptui offline <server>… --yes and spinuptui online <server>… --yes, plus --all for every server and --site <domain> for one site. Without --yes they print the plan and change nothing. spinuptui offline on its own lists what's offline. Servers must be named exactly (name, IP or id); a partial name is refused with suggestions.
  • No alert flood. With Uptime Kuma connected, the affected sites' monitors go into one maintenance window per server. Sites leave it as they come back, and the window is removed when the last one does.
  • You can see what's down. Offline sites show ⏻ off in the site list. Servers show ⏻ off when all their sites are offline and ⏻ 3 off when some are, and they're listed under the Overview's "Needs attention".

Notes

  • Offline means a connection error, not a maintenance page. Visitors see their browser's "can't connect" message. Disabling or re-enabling a site takes a few seconds; SpinupWP works through one server's sites in turn, so a server with 20 sites takes a minute or two.
  • The site's PHP-FPM pool and cron keep running. The pool only listens on a local socket, so it isn't reachable from outside.
  • Details in docs/taking-sites-offline.md.

View on GitHub β†’

v1.1.2

ssh-exec allows quoted wp flags; Kuma errors show their cause

October 15, 2026

A small fix release. spinuptui ssh-exec stops refusing a harmless WP-CLI read, and a failed Uptime Kuma connection now tells you why it failed.

Fixed

  • A quoted wp command that ends in a flag is allowed. spinuptui ssh-exec <site> -- bash -c 'wp --info' was denied because the closing quote landed on the flag and the guard read it as an option missing its value.
  • Uptime Kuma connection errors show the actual reason. With a long instance URL, the cause (a TLS, DNS or proxy problem, say) ran past the edge of the connect form, leaving only Couldn't reach Uptime Kuma at https://…:. The error lines in the connect form and the Alerts panel now wrap. If you couldn't connect before, update and try again: the message should now say what went wrong.

View on GitHub β†’

v1.1.1

ssh-exec's read-only guard closes its Redis and WP-CLI gaps

October 1, 2026

A fix release for spinuptui ssh-exec, the command that lets an agent run diagnostics on a server without being able to change it. Its read-only guard let Redis writes and many WP-CLI writes through. Both are closed, and the update command now reliably fetches the newest version.

Fixed

  • Redis writes are denied. The guard had no rules for Redis, so redis-cli CONFIG SET …, FLUSHALL, FLUSHDB, BGSAVE and plain SET/DEL ran as if they were diagnostics. redis-cli is now allowed only with a known read: ping, info, dbsize, lastsave, config get, memory doctor|stats|usage, latency latest|history, slowlog get|len, client list|info and a few key-inspection reads. Everything else is denied, including unrecognised options and a bare redis-cli fed from stdin.
  • Unlisted WP-CLI writes are denied. The guard named a dozen WP-CLI writes and allowed every other command, so wp transient delete, wp search-replace, wp eval, wp plugin uninstall and wp cron event run got through. wp is now allowed only with a known read (plugin list, option get, core version, cache type, cron event list, db size and the like), and --exec, --require, --ssh and --http are refused on any command.
  • Denials say what was refused. The message names the verb or option, for example redis-cli `config set` is not on the read-only allowlist.
  • The update command now actually updates. The About panel (?) suggests bun add -g spinuptui@latest --no-cache. Without --no-cache, Bun can resolve @latest from a cached copy of the package listing and reinstall the version you already have.

Changed

  • wp user list, wp user get and wp user list-caps are allowed. The guard used to refuse everything under wp user, reads included. Creating, deleting and changing users is still denied.

Notes

  • Two reads that used to pass are now denied: wp config get and wp config list, because they print the database password into the audit log.
  • A WP-CLI read that isn't on the list is denied too, where it used to pass. If a diagnostic you rely on is refused, open an issue and it can be added.
  • The guard still isn't a sandbox. Outside redis-cli and wp it matches known-destructive patterns, so it catches mistakes, not a determined write.

View on GitHub β†’

v1.1.0

Clones match their source's SpinupWP settings, plus `spinuptui siblings`

September 30, 2026

A cloned site now matches its source's SpinupWP settings, not just its files and database, and the deploy script puts itself back. SpinupWP's new site-settings API made both possible. There's also a new CLI command for problems that belong to the whole server rather than one site: spinuptui siblings lists everyone else on the box.

Highlights

  • Clones match the source's SpinupWP settings. Creating a site through SpinupWP's API only sets page caching on or off, so a clone used to get SpinupWP's defaults for everything else. The wizard now copies:

    • page cache duration and exclusion lists,
    • the security toggles (PHP in uploads, xmlrpc.php, multisite subdirectory rewrites),
    • path redirects,
    • the WP cron interval, read from the site's crontab since the API doesn't report it.

    Only what differs is written, so a retry changes nothing, and nothing the clone already has is removed. The verify pane (v) lists what it matched and anything that didn't take, and w opens the new site in SpinupWP to finish it.

  • The deploy script puts itself back. SpinupWP clears a deploy script set when a site is created. Now that its API can update a site's Git settings, the wizard sets the source's script again once the site is done, and fixes push-to-deploy the same way. The g copy-and-open fallback from 1.0 only appears if the script still doesn't stick.

  • Basic auth is flagged. SpinupWP never reveals a site's basic-auth password, so a password-protected source can't be copied. Verify says so, and w opens the new site to set it.

  • spinuptui siblings <domain> β€” who else is on this server. Prints every site sharing a server with the one you name, as JSON, with the server's IP, SSH port, size and Ubuntu version. For each site it shows whether it's the server's vanity host, WordPress or not, its PHP version, whether page cache is on, and whether SpinupTUI set up Uptime Kuma monitors for it. It's for symptoms that belong to the whole box: a vanity host's ?healthz trips on whatever load the busiest co-tenant is making, and site isolation means the vanity site's own user can't see any of it. --probe also SSHes into each site user to show whose logs this device can actually read. A site that comes back permission_denied is one key grant (K) away.

Fixed

  • spinuptui ssh reports a hung connection as connection_failed, instead of an unexplained ssh_error, when its connectivity probe runs past 15 seconds.

Notes

  • Backups don't carry over to a clone yet. SpinupWP stores backups per domain, so a clone backing up before DNS moves would write into the production site's folder. Switch backups on for the new site after cutover; having the wizard do it at the cutover step is planned.

View on GitHub β†’

v1.0.0

Homebrew install, Radicle clones, and your cron jobs come along

September 24, 2026

SpinupTUI reaches 1.0. It installs with one Homebrew command, the clone wizard now handles Radicle sites and brings your cron jobs along, and the header shows how much of your API budget is left before you run into it. From here on, SemVer applies properly: breaking changes to commands, flags or the config file mean a major version.

Highlights

  • Install with Homebrew. brew install mwender/tap/spinuptui installs SpinupTUI and Bun together, and brew upgrade spinuptui keeps it current. The About panel (?) recognizes a Homebrew install and shows that command. Installs use the exact dependency versions this release was tested with.
  • Clone Radicle sites. Git-deployed Radicle sites now clone properly:
    • Uploads come across from content/uploads.
    • After the database import, the site's own deploy script runs on the new server, so the front-end is built (npm ci, npm run build, Acorn caches) just as a normal deploy would build it.
    • Before creating anything, the wizard stops if the site has no deploy script, or if its script needs Node and the destination doesn't have it. SpinupWP servers don't come with Node, so install it on the destination first.
  • Clones keep your cron jobs. Lines you added to a site's crontab (a feed import, a report, an Acorn or Laravel command) are copied to the clone under a # Carried over from … comment. Commented-out lines stay commented, SpinupWP's own WordPress cron entry is left to SpinupWP, and a retry never adds a line twice.
  • Clones match the source's settings. Page cache, push-to-deploy and the deploy script now come from the source site instead of fixed defaults.
  • API headroom in the header. API 52/60 shows how many SpinupWP API requests are left this minute. It's dim while there's plenty, amber once three-quarters are spent, and red when SpinupTUI starts spacing out requests, so a slow moment has a visible reason.

Fixed

  • Plugins & themes (e) no longer shows an empty list on sites with noisy plugins. Some plugins print PHP notices into wp-cli's output, sometimes on the same line as the plugin list. The list is now found wherever it sits, and if wp-cli returns no list at all, the view says so and shows what it printed instead. The WordPress core update check uses the same parsing.

Notes

  • SpinupWP clears a deploy script set through its API a minute or two after a site is created, and there's no API call to set it afterwards. After a clone, the wizard checks the new site. If the script is gone, the roster marks the site, and in the verify pane (v) g copies the source's script and opens the new site's Git settings so you can paste it back. The same check flags a push-to-deploy setting that didn't carry over.

View on GitHub β†’

v0.28.0

Multiple SpinupWP accounts

September 23, 2026

If you look after sites in more than one SpinupWP account, say your agency's own and a client who owns theirs, you no longer need separate setups or token juggling. Keep them all in SpinupTUI and switch with one key, always working in exactly one account at a time.

Highlights

  • A opens Accounts, from any tab. Add an account, switch between accounts, rename them, or remove one:
    • A new account's token is checked with SpinupWP before it's saved, and the account slug (used by w) is pre-filled from the name.
    • A help line under the list says what ⏎, e and x will do to the selected row.
  • One account at a time, on purpose. There's no merged fleet view, so a reboot or DNS change meant for one client can't land on another's server. The header always shows Account: <name> A switch.
  • Switching is a clean reload. Servers, sites, events and probes are all fetched fresh for the new account. If a change is still running (a PHP upgrade, DNS edit, database backup, clone, WordPress update…), the switch waits and tells you what's still going.
  • Each account keeps its own settings:
    • Per account: its token, DNS provider connections, local-copy links, sudo users and saved Keychain passwords, Uptime Kuma monitor mappings, and cached probes.
    • Shared by all accounts: the terminal app, local scan roots, and your Uptime Kuma login.
  • Removing an account forgets it on this machine: its token, Keychain passwords and caches. Nothing changes in SpinupWP, and your local working-copy files stay.
  • Vanity Site in Stacks. Non-WP gains a Vanity Site row for each server's own-hostname site. It's known from the API alone, so these no longer sit under Unknown or unprobed waiting for an SSH probe.

Notes

  • Your existing setup becomes the first account automatically. It's labeled with its account slug until you rename it (A β†’ e). Its Keychain entries and cache files keep their names, so nothing moves and you won't be asked for sudo passwords again.
  • Back up before upgrading if you might downgrade. The first time this version saves a setting, config.json is rewritten in the new shape, and older versions can't read the token from it.
  • Accounts are stored per machine. Add them on each machine where you use SpinupTUI.
  • Commands: spinuptui where lists your accounts, and SPINUPTUI_ACCOUNT=<id> picks one for a single command. A SPINUPWP_ACCESS_TOKEN in the environment pins the account, and the overlay tells you so.
  • Full details: docs/multiple-accounts.md

View on GitHub β†’

v0.27.0

Update WordPress core, one site or a whole version group

September 23, 2026

When a WordPress security release lands, you usually want it on every site now, not after a round of manual logins. This release makes that a couple of keystrokes: update core on one site, or on every site running a given version, with a canary site first.

Highlights

  • u is now Update. It opens a menu with PHP version (the flow you know) and WordPress core. The WordPress option checks the site over SSH with wp-cli and lists the versions you can move to, with the newest minor (security) release picked first. After you confirm, it runs wp core update, then the database upgrade (network-wide on multisite), then reads the version back to verify. It keeps running if you close the overlay, and a toast reports when it's done.
  • Stacks groups WordPress sites by core version. Under Standard WP, Bedrock and Radicle, one row per version (β”” WP 7.0.2 51) shows who is still behind. The newest version in the fleet is highlighted.
  • Update a whole version group. Press u on a version row:
    • It checks every site live first, because cached versions go stale as soon as sites update themselves.
    • Then it shows a plan you can edit. The default target is the newest release in each site's own line (the low-risk security hotfix); m switches to the newest release overall, and space leaves a site out.
    • A canary site always updates alone first; if it fails, nothing else is touched.
    • The rest follow up to 4 at a time, never two on the same server, and the run stops after 3 failures. x stops starting new sites.
  • Progress shows on the rows. A site that's updating shows β § β†’ WP 7.1.2, a failure leaves ⬆!, and sites move into their new version row as they finish.
  • Bedrock and Radicle are included, with a warning. The update patches the files on the server, but composer.lock still pins the old version, so the next composer install (a deploy, a clone) reverts it. The confirm screen names the pinned version, and bulk plans mark those sites βš™. Run composer update and commit the lock to make it stick.

Fixed

  • SpinupWP API errors say why. A rejected write used to show only "SpinupWP API error (HTTP 400)." It now includes the API's reason, such as which field it rejected.
  • Version checks survive noisy plugins. Some plugins print PHP deprecation notices to stdout under newer PHP. That output buried wp-cli's update list, so a site two versions behind read as up to date. Core commands now run with plugins and themes unloaded, which also means a plugin that fatals can't block the core update that fixes it.
  • Row versions stay current. Any version read by a check or an update now refreshes the version cached from the last d probe.
  • Failure toasts are red. They used to show a green βœ“.

Notes

  • u now opens a menu first. Press p (or Enter) to reach the PHP picker you're used to.
  • WordPress core updates use your SSH access as the site user, not the API token, so they work with a read-only token.

View on GitHub β†’