Command palette: rebuild the nav list from a role-gated table so every
navigable page (Docker, Terminal, WireGuard, Commands, Security posture,
Updates, Certificates, Uptime, Alerts, Threats) appears with a role-gated
"g <key>" chord. The help panel is derived from the resulting list, so a
gated-out page has neither an entry nor a live shortcut (no 403 dead-end).
New chords k/t/w/c/p/u/r/a resolve via nav.find; no collision with the
hardcoded CMDK_GO map (d/s/i/f/l/e/y/v).
Scrollbars: add a .no-scrollbar utility and apply it to the server-detail
tab strip and the sidebar nav, so macOS "always show scroll bars" no longer
paints a permanent track where nothing overflows.
Threats KPIs: replace the orphaned 5-box row with a 4-box even grid of the
count metrics + a full-width Top-Attacker spotlight card (an IP, not a count).
serviceAction: a start/restart can EXIT 0 (systemd accepted the request) yet
the unit dies on launch (e.g. monit crashing on a bad config) -> is-active
reports failed. Verify the real post-action state and, when the unit did not
come up, report failure WITH the reason (systemctl status tail) so the
operator sees why instead of a false "gestartet". Harden all three systemctl
calls against argument injection: unit must start alphanumeric, passed after
`--` and escapeshellarg'd (Codex-reviewed, clean).
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Audit leak-hygiene findings:
- The operator's real internal dev-VM/WireGuard IPs (10.10.90.136/.162/.165) were shipped in
vite.config.js (hmr.host), lang hint/placeholder strings, a code comment and the CHANGELOG.
vite HMR host is now env-driven (VITE_HMR_HOST, dev .env only); all other occurrences are replaced
with RFC5737 documentation addresses (203.0.113.x). Tests updated in lockstep.
- docs/ is export-ignored from the git tree and named in promote.sh's refuse-to-publish list, but was
never .dockerignore'd, so `COPY . .` baked the private docs (with private host/URL refs) into the
public image, which the git-tree leak-guard never inspects. Added docs/ (and art/) to .dockerignore.
- /update.log (new repo-root update transcript) gitignored.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Opening a binary / non-UTF-8 file in the file manager threw "undefined is not
valid JSON". The editor bound the raw file bytes to its public $content
property; Livewire JSON-encodes the component snapshot, and invalid UTF-8 makes
PHP's json_encode return false, so the client received an empty response and
JSON.parse("undefined") failed (the trace in mergeNewSnapshot/deepClone).
Fixed at both layers: FleetService::readFile blanks the content when it detects
binary/non-UTF-8 (returns the binary flag instead), and FileEditor::load only
binds content for non-binary, non-oversize files. Binary/oversize files show
their existing notice; the snapshot is always valid UTF-8.
Tests: load blanks binary content (keeps the snapshot valid) and keeps valid
text content. 374 pass, Pint clean.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Journal was a one-time snapshot mislabeled "journalctl -f": it loaded the last
N lines once and never refreshed, so it was neither live nor complete.
- FleetService: `journalctl --show-cursor` + new journalSince() reads only entries
after the stored journald cursor (no gaps, no dup appends), capped to a bounded
live tail; parseJournal split into splitJournalCursor + parseJournalLines.
- Services\Index: pollJournal() (wire:poll.5s) appends new entries since the cursor
and caps retained rows at 300 (fetch ceiling == display ceiling).
- Honest labels: drop the misleading "-f"; badge stays "live" (now true).
Command palette (Strg/⌘ K) can now jump to a server: the fleet feeds the Alpine
cmdk x-data; servers surface only while searching, matched on name OR IP, and
selecting one navigates to /servers/<uuid>.
Cursor is validated (^[a-z0-9;=]+$) and single-quoted before the remote shell —
no injection (Codex-reviewed: no P1, no security findings). Removes stock
ExampleTest (asserted 200 on an auth-gated route → always 302).
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Auth success alone does not prove the server can run commands (a server that
authenticates but rejects exec channels would be accepted, yet every fleet
operation runs commands). The probe now echoes a unique marker and requires it
in the output — proving exec works, while still tolerating servers that omit
the optional exit-status packet.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
testConnection gated success on `exit code === 0`, but phpseclib returns false
from getExitStatus() when a server omits the optional exit-status message, which
would reject a valid login with an empty error. A successful connect() already
proves the credential (it throws on auth failure), so the probe now succeeds on
any exception-free connect + exec.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Connect + trivial exec against a server's stored credential; never throws,
returns {ok,error}. Used by the create-server guard to verify the SSH login
before persisting.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Rework the server-hardening UX per operator feedback ("only activate, can't
deactivate or adjust, UFW won't even activate, is the existing state read?").
- Bidirectional toggles: each item (SSH root-login, SSH password-login, fail2ban,
UFW, unattended-upgrades) is a clean Aktivieren/Deaktivieren toggle driven by the
live feature state, not a one-shot "Anwenden".
- Real state read (HardeningService::state, privileged): SSH from `sshd -T` (the
EFFECTIVE config — honours Include order + Match blocks), UFW from `ufw status`,
unattended from `apt-config dump` (effective periodic value), packages via dpkg —
so nicht-installiert / inaktiv / aktiv are detected; a feature counts as "secure"
only when installed AND active.
- UFW activation installs ufw if missing (fixes "ufw: not found"); opens the detected
sshd port + 80/443 before `ufw --force enable`; adds disable().
- SSH drop-in renamed 00-clusev.conf (sorts + wins first); apt periodic written to a
last-winning 99zz-clusev file. Long timeout for apt installs.
- preview() returns the EXACT command apply() runs (single source — no drift between
the confirmation preview and the executed mutation). Password-disable lock-out guard
intact (refused when Clusev uses password auth or no key exists).
R15 — Codex review gate: `codex review --uncommitted` run iteratively; fixed every
finding across 5 rounds (drop-in precedence, apt periodic config, preview/apply drift,
ufw status detection, privileged-read error handling, stale-config secure state) until
"no actionable regressions". 0 security issues throughout.
Live-verified on 10.10.90.162: state via sshd -T (effective), fail2ban + unattended
toggled on/off both ways, ufw installed + state reflects, lock-out guard refuses
disabling password auth. R12: server detail 200 / 0 console errors, toggles render.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
- fix(hardening): fail2ban/unattended apt installs failed because the 12s SSH timeout
aborted the package download. runPrivileged/runPlain take a $timeout; apt actions use
600s. Live-verified: fail2ban installs + is active/enabled on 10.10.90.162.
- feat(release): cut the foundation as the first real semantic release v0.1.0 (no more
-dev). CHANGELOG is Keep-a-Changelog BY VERSION; the Versions page parses it by version,
reads the newest git TAG as "latest release", and honestly compares installed vs latest
in the channel. Channels reduced to stable|beta (no user-facing dev); System clamps a
legacy value back to stable. Tagged v0.1.0.
- fix(domain/.env): the panel domain lives in the DATABASE (Settings); the app never
rewrites .env. AppServiceProvider overrides config('app.url') at runtime from the Setting
(guarded against a missing table); /system states this explicitly. Caddy config is a
standalone generated file; a Caddy reload is the only infra step.
- docs(arch): Caddy is the prod-only TLS reverse proxy IN FRONT of the app container's
nginx (Internet -> Caddy:443 -> nginx:80 -> php-fpm); no Caddy in dev.
- R15 (rules.md + CLAUDE.md): Codex must review every change for errors + security before
"done". The Codex CLI is not installed in this runtime (needs `sudo npm i -g @openai/codex`
+ `/codex:setup`); an independent adversarial security review of this diff returned CLEAN
(no vulnerabilities, no bugs) as the interim gate.
Verified (R12): /system + /versions 200 / 0 console errors; channels stable|beta only,
DB-not-.env callout; versions by-version with v0.1.0; app.url override resolves from the
Setting; fail2ban live install ok.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Security pivot — make server hardening, SSH access, firewall, dashboard domain/TLS
and the release channel controllable from the dashboard (no SSH needed), with guards
so a remote change can never lock the operator out.
Foundation
- ssh_credentials gains name + disabled_at + last_used_at; CredentialVault refuses a
disabled credential. New key/value Setting model. FleetService::runPrivileged() /
runPlain() — central sudo-aware exec (base64-wrapped sh -c), live-verified as root.
A · control-plane self-hardening
- SecurityHeaders middleware: env-aware CSP (allows the Vite dev origin in dev, strict
same-origin in prod), X-Frame-Options DENY, nosniff, Referrer/Permissions-Policy,
HSTS when secure. 2FA brute-force throttle (5/60s). install.sh sets
SESSION_SAME_SITE=strict + EXPIRE_ON_CLOSE + SECURE_COOKIE (only behind TLS).
B · SSH credential management
- name/label on the access; a credential card on the server page with Bearbeiten /
Sperren-Entsperren (kill-switch) / Löschen (R5), all audited.
C · server hardening from the dashboard (guards + confirmation)
- HardeningService (PermitRootLogin no, PasswordAuthentication no, fail2ban,
unattended-upgrades) + FirewallService (UFW). HardeningAction modal previews the
exact root commands before applying. GUARDS: refuse to disable password-login when
Clusev itself logs in by password or no key exists; UFW opens the real sshd port +
80/443 before enabling. Live-verified non-destructively (previews, the password
guard refusing, ufw status read).
D+E · System page (/system)
- Dashboard Domain + Let's-Encrypt email (Setting) -> DeploymentService renders the
matching Caddy site block (honest: stages the file + shows the reload command, never
fakes TLS). Release channel (stable|beta|dev) configurable; Versions reads it.
Built largely by 4 parallel agents into disjoint files; shared files integrated + the
security-critical bits hardened by hand (the password-auth lock-out guard + env-aware CSP).
Verified (R12): /system + server detail + all 8 routes 200 / 0 console errors (CSP does
not break Livewire/Alpine/Vite); credential card + 5 hardening "Anwenden" buttons render;
the hardening modal opens with the command preview; System persists domain/channel +
renders valid Caddy config; runPrivileged runs as root + the vault refuses disabled creds.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
- Service control now works with a sudo-PASSWORD credential (not only NOPASSWD):
FleetService::sudoFn() defines a remote priv() that feeds the vaulted password to
`sudo -S` over stdin (base64 over the encrypted channel, never on the argv) when not
root, else falls back to sudo -n. Verified LIVE on 10.10.90.162: cron restart/stop/
start executed for real (ActiveEnterTimestamp advanced to now, is-active=inactive
after stop, restarted clean), journal reads real entries via sudo.
- Login brand panel: removed fabricated telemetry — fake "24 Hosts erreichbar",
"cpu 31% mem 48% load 0.86", "clusev connect 10.10.90.0/24". Replaced with true
capability lines (agentless SSH/phpseclib, TOFU host-key pinning, 2FA, audit log,
AGPL). Only real claims now (per "nur eintragen was auch wirklich geht").
- New Version & Releases page (/versions, EN route per R13) with REAL data only:
version from config/clusev.php, build SHA + branch read from .git at runtime,
changelog parsed from a real CHANGELOG.md, real Gitea repo + AGPL license, honest
update path (deploy commands) — NO fake updater / stars / forks / CVEs / "update
available v2.5.0". Sidebar nav + tag/git-branch icons added.
- Dummy-data sweep (12-auditor workflow + adversarial verify): 1 confirmed finding —
removed the unused sine/cosine series() fake-sparkline generator in Dashboard.php;
also fixed a stale "static seed" chart comment and a stale "mock listing" comment;
derived the versions repo label from config (DRY). 5 false positives dismissed.
Verified (R12): /login + /versions + all 7 routes 200 / 0 console errors; login
fake-metrics gone + honest lines + fonts load; versions shows real version/build/
changelog/repo with zero fabricated stats.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Full feedback round — all browser-verified (R12: 7 routes 200, zero console errors)
and functionally tested live against the target host.
- Metrics are real AND honest: poller now stores cpu/mem/DISK/LOAD history (all
four sparklines real, not synthetic); KPIs show absolute context (Memory
"1,9 / 7,8 GB", Disk "13 / 80 GB", Load "4 Kerne"); tiles are status-coloured
and Load is rated against core count (so "yellow" actually means load≈0.7-1.0×
cores), not a fixed warning tint.
- File manager is functional: Hochladen (Livewire upload -> SFTP put), Download
(SFTP get -> streamed), and view/edit (new FileEditor modal: SFTP get, binary/
size guards, SFTP put on save + audit). File names + Bearbeiten open the editor.
- Settings redesigned (/frontend-design): account identity header + section nav
(Profil / Sicherheit) instead of flat stacked panels.
- Routes are English (R13): /einstellungen -> /settings; rule added to rules.md +
CLAUDE.md. Dummy data removed: the topbar "Flotte online / Uptime 42d" + dead
bell button are gone, replaced by a real "<online>/<total> online" pill.
New: app/Livewire/Modals/FileEditor + view; FleetService get/read/write/upload +
Sftp putFromFile/size.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Service start/stop/restart now run over SSH (root direct, else `sudo -n`) and
report the real result instead of an optimistic state flip; the Services list is
reloaded afterwards to reflect the actual unit state. journalctl in systemd() is
likewise run via sudo/root so the full system journal is available once a
privileged credential exists. Unit names are validated before interpolation.
Without a privileged credential the action fails honestly ("sudo: a password is
required") rather than pretending to succeed.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Make the panel actually operate the server, not just display it — for everything
that works in the SSH user's own space (no sudo required).
- FleetService: addAuthorizedKey (append, dedup, perms-safe via base64 transport),
removeAuthorizedKey (by SHA256 fingerprint, preserves every other key — never
rewrites blindly, no lockout), sshKeys (reload), deleteFile (SFTP unlink).
- AddSshKey form modal: paste a public key -> appended over SSH + AuditEvent +
list reload. Wired to the "Schlüssel hinzufügen" button on Server-Details.
- Server-Details: key removal now performs the real SSH removal then reloads;
new "Dateien" button opens the file manager scoped to this server.
- Files: delete performs the real SFTP unlink then reloads the directory.
Verified live: add+remove of a throwaway key leaves the existing key intact;
real file delete confirmed gone. systemctl start/stop/restart still need a
privileged credential (the demo account has no passwordless sudo).
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Addresses "chart/gauges don't update" and "design": the visible live layer no
longer depends on the Reverb WS path (which was unreliable through the dev NAT).
- Poller is now the single SSH metrics source: FleetService::applyMetrics writes
the latest reading + a rolling cpu/mem history into the cache (and the DB).
- Dashboard reads latest + history from cache (no SSH on web render); the big
chart and KPI sparklines render real history; wire:poll.10s refreshes them.
- Server-Details gauges read the poller-updated row via wire:poll.10s; the donut
rings get a size="lg" variant (h-24) so they read clearly + animate.
- FleetSeeder no longer seeds a fake fleet (those servers had no credentials and
showed no data). It seeds one real server from CLUSEV_DEMO_SSH_* env vars, or
an empty fleet otherwise. Existing fake servers removed from the demo DB.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Resolve all 12 confirmed findings from the SSH-integration review.
Security:
- TOFU host-key pinning (new VerifiesHostKey trait + servers.ssh_host_key column):
SshClient/Sftp record the server host key on first connect and refuse the
connection on any later mismatch — phpseclib does not pin host keys itself.
- PollMetrics keeps one long-lived SSH connection per server and reuses it across
ticks (a single login, not one per interval) — avoids fail2ban/auth-log churn;
dead connections are dropped and re-established next tick.
- @js() escaping for every server-controlled value interpolated into wire:click
(file names, paths, service names, SSH-key comments/fingerprints) — prevents
JS-string breakage / injection from untrusted remote data.
Correctness:
- parseKeys regex makes the key comment optional (keys without a comment were
silently dropped).
- parseVolumes pops the three trailing single-token df fields and rejoins the
rest, so mount points containing spaces parse correctly.
- Sftp gains a disconnect() method to match SshClient (explicit cleanup).
Rules:
- Services "Start" -> "Starten" (R9, German).
- Files breadcrumb buttons get min-w-11 + padding (R7, >=44px touch target).
Verified live: host key stored + mismatch refused + correct key reconnects;
poller reuse polls ok; all pages still render real data.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Replace the page mock data with real reads from the target host via the
phpseclib SSH layer. New FleetService parses raw command output into the exact
shapes the Livewire pages already consume; all parsing forces LC_ALL=C and reads
/proc to stay locale-independent.
- app/Services/FleetService.php: metrics (cpu via two /proc/stat samples, mem via
/proc/meminfo, disk via df), full snapshot (identity/specs incl. virt+disk_gb,
volumes, interfaces from ip+/proc/net/dev, sshd/fail2ban/ufw hardening, authorized
keys), systemd units + journal, and an ls-based directory listing. One compound
command per read; connect/parse failures bubble up.
- clusev:poll-metrics command replaces the mock emitter in the dev supervisor:
polls every credentialed server, persists cpu/mem/disk/status, broadcasts
MetricsTicked(server) — unreachable boxes flagged offline, loop never dies.
- Pages wired with graceful failure (offline state, never a 500):
Dashboard (live cached metrics + notable units), Services (real units+journal),
Files (real listing + dir navigation via open/go/up), Server-Details (live
snapshot persisted onto the row + offline banner).
- WithFleetContext prefers a credentialed, non-offline server as the default.
- dualChart filters ticks by server name so the chart tracks the active host.
Verified live against a real Debian 13 box: metrics/services(91)/journal(25)/
files(navigable)/snapshot(5 ifaces, real hardening, real key) all parse correctly.
Credentials are stored encrypted in the vault — never in source.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
boban