Fixes from the security audit (the brute-force/rate-limiting angle matters more now that the
password policy is min(6)/no-complexity, so IP-keyed throttles must not be spoofable):
- bootstrap/app.php: trustProxies(at: '*') → trust only PRIVATE ranges in prod. Caddy reaches
app:80 from the docker bridge (private IP) so it stays trusted, but a public/off-network source
can no longer forge X-Forwarded-For to spoof request()->ip() and bypass the login/2FA/forgot
throttles + the brute-force ban.
- TerminalSession + HostCredential: replace $guarded=[] with an explicit $fillable allowlist; add
$hidden=[secret,passphrase] to HostCredential so it never serializes its credential fields.
- nginx /terminal/ws: access_log off — the single-use session token rode the query string into the
access log.
Verified: full suite 467 pass; all panel pages 200 with zero console errors; server terminal still
connects + runs commands.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Operator decision: password strength is the user's own responsibility. Removed the
mixedCase + numbers requirement from every password rule (Password::min(6) is all that
remains), so a plain 6-char password ("abcdef") is accepted. Dropped the
"upper/lowercase + a digit" clause from the hints (de + en).
Verified: a 6-char simple password validates; a 5-char one still rejects (length); full
suite 467 pass.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Operator decision — 12 was too long; 6 is acceptable and password strength is the user's
own responsibility. Changed Password::min(12) → min(6) in every place that validates a
password (password change, forgot/reset, settings profile, create-user) and updated the
visible hints (de + en). The upper/lowercase + digit complexity is unchanged — only the
minimum length was flagged.
Verified: a 6-char complex password ("Abc123") now validates; too-short and complexity-less
passwords still reject; the password page shows "min. 6"; full suite 467 pass.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
The Add-account modal only ever generated a one-time password — there was no way to set
one. Add an optional password field (with a show/hide toggle):
- blank → previous behaviour: a strong one-time password is generated, the account is
flagged must_change_password, and the clear text is revealed once.
- filled → that becomes the user's own login (validated min 12 / mixed case / digit);
must_change_password stays false (no forced rotation) and nothing is revealed.
Browser-verified: the modal shows the password field + hint; both paths covered by tests.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
- sidebar: the "2FA on/off" text badge next to the user name was visually noisy. Replace it
with a single status shield — a cyan shield-check when 2FA is on, a muted struck-through
shield when off; the label moves to the hover/aria title so it stays reachable.
- terminal: the server-target search now appears once there is more than one server (was 5+),
so it is there as soon as filtering helps; added a clear (×) button + an accent focus ring.
Browser-verified: sidebar shows just the shield (no text); search filters the rail (homelab →
only homelab-01) with a working clear button; both badge states render correctly. 465 pass.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
The stored secret is never echoed back (security), so on reopen the blank password field
read as "not configured" / "it forgot my password". The login DOES persist (single
HostCredential row, kept across reloads); only the modal was ambiguous.
- when configured, show a green status: "Configured as <user> — password is stored. Leave
blank to keep it; enter a new one to change it." The Remove button + gear stay, so the
operator can still change the password or switch access.
No behaviour change — set once, stays until changed/removed.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
The "Clusev host" terminal is for the LOCAL machine Clusev runs on. Letting the operator
point it at an arbitrary host/server makes no sense (that is what fleet servers are for)
and the pre-filled host.docker.internal read as a confusing mandatory magic value.
- the host is now FIXED to host.docker.internal server-side (HostShell::HOST), not a form
field; the modal asks only for the login — user + password|key, and the port.
- the rail tile sub-label shows just the login user (the host is implied/local).
Browser-verified: modal has no HOST/IP field (user/port/auth/password only); saving shows
the user on the tile; opening the host reaches the real host sshd (a wrong password yields
an SSH auth error, proving the connection), zero console errors. 16 terminal tests pass.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
The field is pre-filled with host.docker.internal — a Docker DNS name that resolves to
the machine the containers run on (the Clusev host) — but had no explanation, so it read
as a mandatory magic value. Add a one-line helper under the field: it's the default for
the Clusev host; leave it, or enter another IP/host. The field stays editable on purpose
(some setups want the host's LAN IP or a different machine).
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
The seeded-password rotation was forced before the panel was reachable. It is now a
nudge, not a wall: the prompt shows once per session and can be skipped, and a standing
warning banner keeps reminding while the default password is still in use.
- EnsureSecurityOnboarded redirects to the password page only when must_change_password
AND the session has not set onboarding.password_skipped (set by PasswordChange::skip,
which lands on the dashboard WITHOUT rotating — so the flag/banner persist).
- app layout shows a default-password warning banner while must_change_password is true.
- the metrics broadcast channel no longer gates on securityOnboarded() — a deliberately
un-rotated (but authenticated) operator would otherwise lose realtime metrics;
/broadcasting/auth still runs in the web group (guest = 403), so it stays auth-gated.
- 2FA was already optional (never enforced) — only the login copy was wrong: "2FA
required/enforced" → "recommended/optional" (de + en).
Browser-verified (R12): login → skippable prompt → dashboard banner → free navigation
(no re-redirect); login copy correct; zero console errors. Full suite 465 pass.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
The "Clusev host" terminal was a shell inside the sidecar container (node@clusev, no
sudo, not the host). It is now a REAL SSH login into the host machine, like any fleet
server — so it shows the real root@<host> prompt with full rights.
- HostCredential (encrypted singleton) holds the host SSH login; a HostShell modal lets
the operator set host/port/user + password|key. The clear-text secret is never echoed
back; on edit a blank secret keeps the stored one, and switching auth method requires a
fresh secret (no password silently reused as a key).
- open('host') mints a kind=host token only when a login is configured, else opens the
setup modal; the internal resolve endpoint returns a server-shaped SSH spec (decrypted,
internal-net only) for kind=host, or 404 when unconfigured.
- compose: extra_hosts host.docker.internal:host-gateway so the sidecar reaches the host
sshd; Caddy now 404s /_internal/* at the public edge (the sidecar uses app:80 directly).
- rail gains a debounced search box past 5 servers; removed the now-obvious PTY hint line.
Browser-verified (R12): host tile shows "not configured" + gear → modal → SSH login as
root runs commands, zero console errors; search filters; resolve returns the host spec.
15 terminal tests pass.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
GitHub deprecated the Node 20 action runtime, so checkout@v4 and setup-node@v4 were
force-run on Node 24 with a warning. v5 of both targets Node 24 natively; their inputs
are unchanged (ref/repository/token/fetch-depth/path; the build's node-version stays
'20', that's the project runtime not the action runtime). setup-php@v2 is the current
major and unaffected. ci-staging runs only on tag pushes, so the warning clears on the
next tagged release.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
On a phone the terminal clipped on the right: FitAddon measured the column count before
JetBrains Mono finished loading (narrower fallback metrics → too many columns → the wider real
font then overflowed and the right edge was cut). And the PTY opened at a fixed 80x24, so the
remote shell's first screenful (MOTD/prompt) was generated at 80 columns and wrapped awkwardly
on a ~43-column viewport.
- refit immediately, on the next animation frame, and again on document.fonts.ready (font race)
- refit on the 'ready' frame before sending the size
- pass the browser's cols/rows on the WS URL; the sidecar opens the host PTY / SSH shell at that
size (clamped 1..1000), so output is born at the visible width
Verified in-browser at 390px: PTY reports 43x23 (matches the box), no horizontal overflow, clean
wrapping, zero console errors; desktop server+host terminals and cross-origin rejection still pass.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
The host terminal is a shell in the sidecar container; it opened in the read-only
repo mount (/workspace) with the prompt showing the random container id, so it read
like a stray container rather than "the Clusev host".
- give the sidecar a stable `hostname: clusev` (dev + prod) → prompt host is `clusev`
- land the host PTY in /home/node (a real, writable home) and set HOME to it, so the
shell opens at ~ like a normal terminal instead of /workspace
Server (SSH) terminals are unchanged. Verified in-browser: prompt reads node@clusev,
pwd is /home/node, HOME=/home/node, no container hash / no /workspace; server terminal
still connects and runs; zero console errors.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
A dedicated /terminal page with a target rail: one terminal per fleet server (SSH)
plus one for the Clusev host itself. php-fpm cannot hold an interactive PTY, so a
small Node sidecar (ws + ssh2 + node-pty) bridges xterm.js to the PTY.
Flow: the Livewire page mints a single-use, 60s TerminalSession token and dispatches
it to an Alpine xterm island; the browser opens a same-origin WS (/terminal/ws,
proxied by nginx in dev / Caddy in prod to terminal:3000); the sidecar burns the
token at an internal resolve endpoint (shared-secret header, private network only)
for a connection spec, then opens an SSH PTY (decrypted credential, never sent to the
browser) or a local host PTY. Real shell = native Tab/Shift-Tab completion.
Security:
- token burned ATOMICALLY (single conditional UPDATE) — no double-open race
- resolve endpoint guarded by hash_equals shared secret, exempt from CSRF + PanelScheme
host enforcement (private net only), returns decrypted creds over the internal net
- sidecar enforces same-origin on the WS upgrade (hostname match, port-agnostic)
- single-use 60s tokens, swept daily so the table can't grow unbounded
Verified in-browser (R12): host + server terminals connect, accept keyboard input,
run commands, Tab-complete; cross-origin WS rejected; zero console errors. 9 feature
tests cover minting, the resolve spec, single-use burn, expiry, and locked creds.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
The series rail synced ?series=<key> via Livewire #[Url] with a bare numeric key like
"0.10". Livewire type-juggles that to 0.1 (the trailing zero is lost), so reloading
while viewing the 0.10 series silently switched to 0.1 and rewrote the URL. Prefix the
series key with a non-numeric "v" (?series=v0.10) so it round-trips as a string; the
rail label still shows the dotted "0.10". Surfaced now that a 0.10 series exists.
Browser-verified: selecting 0.10 -> ?series=v0.10, and it survives a reload.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Server-Details resource-history graph (CPU/RAM/Disk over time, selectable range +
hover tooltip, range persisted in the URL), nameable SSH keys, and two hardening
fixes (SSH-root-login self-lockout guard, fail2ban systemd-journal backend). See
CHANGELOG.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Clicking a range now writes ?range=<key> (replaceState — no history spam) and the
chart restores it on init, so reloading the SAME page keeps the chosen range. Opening
the page fresh (no param) still defaults to 24h.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
The y-axis labels spanned the whole wrapper (inset-y-0), which now included the
x-axis label row — so the "0" tick and the "−1 h"/"jetzt" labels overlapped at the
bottom. Wrap only the plot + y-axis in the relative box and render the x-axis labels
as a separate row beneath it, so "0" sits on the chart's bottom edge and the time
labels are clearly below.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Reworks the Server-Details history graph from a static Livewire-rendered SVG into a
self-contained Alpine island fed by a JSON endpoint, addressing the feedback:
- Smooth Catmull-Rom curves instead of angular segments.
- Continuous line with an area gradient fill; the line breaks ONLY on a real outage
(delta-t > 2.5x the bucket), not on every empty bucket — no more dropouts.
- Hover crosshair + tooltip (time + CPU/MEM/DISK at the point).
- Instant client-side range switching (1h/24h/7d/30d) via
GET /servers/{server}/history.json — no Livewire round-trip.
- X-axis labels moved below the plot (no longer overlapping the 0 gridline).
MetricHistory::series() now returns only non-empty buckets plus from/now/bucket. The
chart wrapper is wire:ignore so the parent's 10s poll never clobbers the Alpine state;
the island self-refreshes every 60s. The old Livewire range logic is removed.
Codex review: clean (auth-gated route, range clamped, numeric-only JSON, guarded hover
bindings, R3/R4-compliant). Browser-verified: smooth curves, area fill, working tooltip
+ crosshair, instant range switch, zero console errors. 446 tests green.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
A CPU/RAM/Disk history graph over selectable ranges (1h/24h/7d/30d) on the server
detail page, backed by a persisted time series (the live 15s view stays cache-only).
- metric_samples table (server_id cascade, cpu/mem/disk %, load, sampled_at indexed).
- clusev:sample-metrics (scheduled every minute, mirrors clusev:wg-sample): persists
one row per server from the cached live reading, prunes beyond --retention (30d).
- MetricHistory::series() buckets + averages samples for a window; empty buckets stay
null so the chart shows gaps, not fabricated values.
- Servers/Show: range selector + a gap-aware SVG line chart (theme-token strokes,
R3-compliant, mirrors the dashboard chart). historyRange is clamped via setRange()
and the updatedHistoryRange() hook (it's a public, client-settable property).
Codex review: fixed a right-edge bucket off-by-one, clamped percentages on write, and
guarded the public range property against direct client values. Browser-verified: the
chart renders with data, range buttons, legend and gaps, zero console errors. 447
tests green, pint clean.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Verified live against a real Debian 12: the hardening actions work end-to-end
(root + password self-lockout guards refuse correctly, auto-updates and ufw enable,
disabling password auth persists and blocks password logins). One gap surfaced: on
a minimal image without rsyslog the distro-default fail2ban sshd jail (logpath
/var/log/auth.log) fails to start — "Have not found any log file". Pin the sshd jail
to `backend = systemd` (reads the journal) via a jail.d drop-in on enable:
log-source-agnostic and valid on every supported (systemd) target. fail2ban then
comes up active with the sshd jail reading the journal.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
The "add SSH key" modal had no name field — keys appeared only by comment +
fingerprint, so additional keys weren't tellable apart. Add a name field; the name
becomes the key's OpenSSH comment (what the server list shows), for both generated
and pasted keys. Sanitised (control chars stripped, whitespace collapsed, capped at
64 chars) so a crafted name cannot inject a second authorized_keys line — proven by
test, on top of addAuthorizedKey()'s own whitespace-collapse + base64 encoding.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Disabling SSH root login (PermitRootLogin no) severs Clusev's OWN access when it
connects AS root — afterwards neither key nor password reaches root. The
password-auth toggle already guards against self-lockout; the root toggle did not.
Refuse to disable root login while the active credential's username is root, with a
clear message: add a non-root sudo user + key and switch Clusev to it first.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
The README banner showed v0.9.40 while the badge showed v0.9.58 — static fields
that go stale and disagree. Removed the per-release version from both the badge row
and the banner SVG (the release version lives in tags/changelog, not a frozen
image). Added a "Supported operating systems" matrix (Debian/Ubuntu · RHEL · SUSE
full; Arch partial; Alpine metrics-only; anything else metrics-only) and real
minimum requirements (CPU/RAM/disk for the panel host; nothing on managed servers).
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Follow-up to the Projekt-card change: the dev/release-control box must list its
actual update source (e.g. private Gitea) AND the public repo — the previous commit
showed only the source on dev. Replaced the single project link with projectLinks():
the public repo is always listed (validated slug, canonical fallback); the dev box
prepends its source when it is a real URL (a garbage/hostless source is dropped, never
an empty link). Staging/stable still list the public repo only — no private leak.
Codex review: no leak path on non-dev; hardened the dev source against a hostless URL.
Verified: 433 tests green (dev-both / non-dev-only / malformed-slug / garbage-source),
pint clean, /versions shows both links on the dev box with zero console errors.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
The Versions "Projekt" card linked config('clusev.repository') — the per-install
update source, which on the dev box is the private Gitea. A staging/stable install
must never surface a private source. Now only the dev/release-control box shows its
actual source; every other install links the public open-core repo only, built from
the public slug with a canonical fallback (a malformed or full-URL slug can never
produce a private/garbage link). The update SOURCE (ReleaseChecker) is unchanged —
this governs only the displayed link.
Codex review: hardened the slug to reject non-owner/repo values; the release_controls
gate is the dev-box invariant (it also gates the dev-only /release page, and a staging
box's CLUSEV_REPOSITORY is the public repo anyway).
Verified: 432 tests green (incl. dev / non-dev / malformed-slug cases), pint clean,
/versions loads 200 with zero console errors.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
The self-update progress screen reset its elapsed timer to 0 and dropped the
phase checklist back to step one on every browser refresh — and the long image
build made it look like the update had restarted. Root causes: the timer and
phase state lived only in page JS (Date.now() at page load), and a
30s-before-page-load staleness guard rejected the real phase feed whenever a
stage had run longer than 30s (which the image build always does).
- DeploymentService: requestUpdate() now also writes a best-effort
run/update-started.json {at,from}. New updateStartedAt()/updatePhase() expose
the server-side start + current macro-stage; updatePhase() requires a positive
timestamp so a corrupt or stale 'done' can never drive a premature finish.
clearUpdateRequest() also removes the start marker.
- routes: /update-status.json delegates to updatePhase(); /update-progress
passes startedAt/initialPhase/hasFeed. A stage is trusted only when tied to
THIS update's start (startedAt) and written at/after it — a leftover stage from
a previous run can no longer resume or trigger a premature completion.
- update-progress.blade.php: the elapsed timer is anchored to the server-side
start (counts the real elapsed across a refresh); the checklist resumes at the
live stage; the live feed is authoritative (corrects any time-heuristic
overshoot) and the heuristic stays off once a real stage is known.
Verified: 28 UpdateProgressTest cases + full suite (426) green, pint clean, and a
headless browser load+reload mid-update keeps the timer counting and the phase in
place with zero console errors. Codex review: two premature-completion edge cases
found and fixed (timestamp validation + start-anchored freshness).
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
The changelog mixed user-facing changes with dev-only maintainer tooling
(the release pipeline, CI fixes), reading as overly detailed/unprofessional.
Establish the convention up front — user-facing changes only; internal work
lives in git history — and apply it:
- Collapse the dev-only release-tooling versions 0.9.59–0.9.68 into a single
terse 0.9.68 "Intern" entry (kept parseable so the Versions page still lists
the running version).
- Trim 0.9.58 to its user-facing parts (GitHub update-check, public address).
- Reduce 0.9.57 to the user-facing release-channels note.
Verified the Versions-page changelog parser still yields clean nodes
(0.9.68 -> Intern, 0.9.58, 0.9.57, ...) with no leaked translation keys.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
- PromotionService::publicTags() is called on every Release render incl. the 5s
page poll; the anonymous GitHub tag API is rate-limited (60/h per IP). Cache it
120s so the panel never gets rate-limited to empty. (Test flushes cache in setUp.)
- yank.yml: replace the loose shell-glob tag guard with an anchored grep regex so
a garbage/space/metachar tag dispatched straight from the GitHub UI is rejected
before reaching git push --delete (the panel already validates tag ∈ publicTags).
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
'Staging' is not a single server: any beta-channel server pulls a release itself,
so a per-server 'deployed' step in the dev dashboard was arbitrary + contradicted
that model. The pipeline now tracks only the global build facts of a beta —
Tag (Gitea) → Mirror (GitHub-private) → CI. Removes PipelineStatus::testStep, the
test-server Setting/input/saveTestServer, and the now-orphaned lang keys.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Full-page tests render @vite layouts; without public/build/manifest.json (gitignored,
absent on a fresh CI checkout) they 500. Move npm ci + npm run build ahead of the
test step so the manifest exists. Fixes the red CI on every v* tag.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
- PipelineStatusTest: the private mirror slug 'boksbc/clusev-staging' was hardcoded
in this tracked test (tests/ ships to the public repo) — exactly the leak the
design forbids. Neutralise to an example slug 'acme/staging'. Add the missing
HTTP-error degrade tests (GitHub 5xx, GitHub throwing, test-server unreachable →
'unknown', never throws) — the core robustness guarantee was untested.
- Live rail: render the per-step description (the built $sub array was dead code)
and colour the live state by status (online/warning/offline).
- saveTestServer: refresh() the pipeline cache so a changed test-server URL is
reflected immediately, not after the 15s TTL.
- .env.example: document the dev-only release/pipeline keys (commented, no values).
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
The GitHub-private staging repo is a full push-mirror of Gitea, so export-ignore
(which only protects the public repo via git archive) does not strip these from
the mirror. Untrack the internal/AI-collaboration docs — CLAUDE.md, rules.md,
handoff.md, kickoff-prompt.md and docs/ (specs, plans, runbooks, handoffs) — and
gitignore them so they stay on the dev box but never reach Gitea, the mirror, or
any shipped image. The public repo was already clean via export-ignore.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
boban