11 Commits

Author	SHA1	Message	Date
Anso	c677b8bb66	fix: fall back to enroll token when pilot tunnel JWT is rejected (#1566) ... * fix: fall back to enroll token when pilot tunnel JWT is rejected On HTTP 401/404 upgrade rejection, delete stale pilot.jwt and retry with SENCHO_ENROLL_TOKEN. * test: import pilot agent module after DATA_DIR is set in fallback test The auth-fallback test statically imported pilot/agent, which freezes its pilot.jwt path from DATA_DIR at module load, before setupTestDb redirects DATA_DIR to a writable temp dir. On the Linux CI runner the path resolved to a non-writable /app/data, so persistToken silently failed and the round-trip assertion read null. Import the module dynamically in beforeAll after setupTestDb, matching the sibling unit test. * fix: remove unexpected-response listener that blocked pilot reconnect The ws library skips abortHandshake when an unexpected-response listener exists, so error and close never fire and the agent hangs in CONNECTING. Detect auth rejection via the abortHandshake error message instead; the close handler already performs enroll-token fallback and reconnect.	2026-07-05 01:03:09 -04:00
Anso	41dc339c26	fix: enforce 1:1 compose path mapping for Pilot agent mounts (#1516) ... * fix: enforce 1:1 compose path mapping for Pilot agent mounts Pilot enrollment now generates validated 1:1 bind mounts so every agent path maps to a unique compose directory. Persisted agent paths reconcile during startup to catch drift. Unsafe relative-bind redeploys are blocked before container removal to prevent path escapes. - Add composePathMapping utility with strict path validation - Generate COMPOSE_DIR and validated mounts during Pilot enrollment - Reconcile persisted agent paths during startup bootstrap - Block redeploy when a relative-bind mount would escape the compose root - Default Pilot UI path to /opt/docker/sencho - Update multi-node and pilot-agent documentation - Add regression tests for enrollment, bootstrap, compose-service, and environment-check paths * fix: update E2E enrollment regexes for YAML-quoted token values	2026-06-29 14:35:50 -04:00
Anso	ce08a593d7	feat(settings): reorganize the settings hub into domain groups (#1321) ... * refactor(settings): split System Limits and regroup the hub System Limits had grown into a grab-bag of host alert thresholds, Docker cleanup, and mesh data-plane controls under one mislabeled section. Split it into Host Alerts, Docker & Storage, and Fleet Mesh, and split Developer into Developer Diagnostics and Data Retention. Reorganize the sidebar into ten domain groups: Personal, Access, Infrastructure, Monitoring, Notifications, Automation, Organization, Security, Operations, Help. Each section now saves only its own keys, so a concurrent edit in one section no longer clobbers another. Data Retention sends the audit-log window only on a paid plan, matching the field's existing visibility, so a Community save no longer fails on a key the operator cannot set. NumberChip moves to a shared module and the toggle reuses the existing shared component. The /settings API is unchanged. * test(settings): cover registry structure and per-section save payloads Add structural invariants for the ten-group registry (every item maps to a real group, ids are unique, the System Limits and Developer splits land in the right groups with the right gates, renamed labels and the Registries paid gate hold) and per-section payload tests asserting each split section patches only its own keys, including the Community path where Data Retention omits the paid audit-log key. * docs(settings): document the regrouped settings hub Rewrite the settings reference for the ten-group layout, replace the System Limits page with Host Alerts, Docker & Storage, and Fleet Mesh, and document the prune-on-update, reclaimable-space banner, and mesh auto-recreate settings that were previously undocumented. Update the Settings navigation breadcrumbs across the feature docs and refresh the affected screenshots. * fix(settings): show Access sections as instance-global, not operator-scoped License, Users, SSO, and API Tokens are instance-global settings but the masthead scope label rendered them as operator-scoped because it keyed off the old Identity group. Only Personal sections (account, appearance) are operator/browser-scoped now; everything else reads as global. Also add a compile-time exhaustiveness guard to the section switch so a future SectionId added without a matching case fails the build instead of silently rendering a blank panel. * docs(settings): remap remaining settings breadcrumbs to the new groups Update the navigation breadcrumbs that still pointed at the removed Identity, Alerts, and Advanced groups: API Tokens and Users now sit under Access, Webhooks under Automation, Labels under Organization, App Store under Infrastructure, Appearance under Personal, and scan policies under Security > Vulnerability Scanning. Correct the settings reference scope note so Access reads as global. * docs(settings): remap renamed-section breadcrumbs across feature docs Sweep every feature, operations, getting-started, and reference page for navigation paths that still named the renamed settings sections, and point them at the current ones: Security becomes Security > Vulnerability Scanning, Notifications becomes Notifications > Channels, Routing becomes Notifications > Notification Routing, and Developer becomes Operations > Developer Diagnostics (with its retention windows under Operations > Data Retention). App Store moves under Infrastructure and the four-group overview in the getting-started intro is rewritten to the ten groups. Separators each page already used are preserved.	2026-06-05 23:01:37 -04:00
Anso	5f7a887ed6	docs: reorganize navigation for feature discoverability (#1147) ... Restructures the Documentation tab so each group answers one operator question. - Split the 12-page "Stacks & Deployments" into Stacks (per-stack work) and Deployment (the act of deploying); promote Resources Hub to a standalone item. - Dissolve the 2-page "Platform" junk drawer: Sidebar moves to Stacks, Host Console moves to Fleet. - Rename "Fleet & Multi-Node" to "Fleet"; absorb Node Compatibility from Reference. Move Scheduled Operations from Fleet to Automation (now a 4-page group covering Scheduled Ops, Auto-Update, Auto-Heal, Webhooks). - Clean up the Reference tab: drop misplaced node-compatibility, move root-level security.mdx into reference/, delete the orphan reference/verifying-images.mdx after porting its Available Tags table into operations/verifying-images.mdx. - Reorder top-level groups: Operations moves above Reference. - Rename two misleading page titles: "Deploy Progress Modal" becomes "Deploy Progress" (drops the UI implementation leak); "Auto-Update Readiness" becomes "Auto-Update Policies" (matches filename and sibling "Auto-Heal Policies"). Verified: docs.json parses as valid JSON, 59 disk .mdx files match 59 nav entries with zero orphans and zero broken refs.	2026-05-21 21:25:23 -04:00
Anso	e65c5e8551	fix(pilot): post-merge audit followups (WS via getProxyTarget, closeTunnel lifecycle, mesh source buffer, docs) (#1128) ... * fix(pilot): route remote WS upgrades through NodeRegistry.getProxyTarget Pilot-mode nodes carry empty api_url and api_token by design and expose their API on a per-tunnel loopback bridge. The upgrade handler gated the remote-forwarder branch on `node.api_url && node.api_token`, so WS requests targeting pilot nodes silently fell through to the local handlers (live logs, exec, generic) instead of tunneling to the agent. Resolve the target via NodeRegistry.getProxyTarget so pilot and proxy modes share one dispatch path, mirroring the HTTP proxy. handleRemoteForwarder now takes the resolved target and, when the target is the pilot loopback (empty token), skips the console-token exchange and the Authorization injection so the tunnel-side auth is the only source of truth on that path. Unresolvable targets reject the upgrade with HTTP 503 instead of being served gateway-local data. * fix(pilot): emit tunnel-down and mark node offline on closeTunnel PilotTunnelManager.closeTunnel closed the underlying WebSocket but skipped the cleanup the natural-disconnect path runs, so explicit closures (enrollment regenerate, node deletion) left the node row at status='online' until the next reconnect. The dashboard kept showing the stale state for the entire interval. closeTunnel now writes nodes.status='offline' and emits tunnel-down for pilot bridges, and emits proxy-bridge-down for central-initiated proxy bridges. The maps are cleared before bridge.close() so the natural 'closed' handler's bridge-identity guard short-circuits and we do not double-emit. * fix(mesh): buffer cross-node source data until tcp_open_ack arrives openCrossNode piped src socket data straight to tcpStream.write before the forward TcpStream emitted 'open'. The first packet on a fresh cross-node stream raced ahead of the agent's tcp_open_ack on the wire, which broke protocols that send immediately after connect (HTTP, TLS, Redis, Postgres) on Pilot and proxy mesh paths. Buffer src chunks in a local array capped at STREAM_PENDING_DATA_MAX_BYTES until tcpStream emits 'open', then flush them in order before any post-open writes. Tear down both sockets if the buffer overflows so a misbehaving source cannot exhaust gateway memory while waiting for the ack. * docs(pilot): clarify host-console non-parity and narrow the parity claim Pilot mode disables the host-console capability at the capability registry (the agent container has no useful host shell to surface), but the public docs listed host console among the WebSockets that ride through the tunnel and described pilot as behaving identically to proxy mode. State the shared-capability claim more carefully and call out the intentional non-parity in a dedicated subsection.	2026-05-21 01:00:47 -04:00
Anso	08caa914ce	docs: v1 docs refresh (batch 2) (#988) ... * docs(atomic-deployments): refresh page around current UI and behavior Rewrites the page to match the v1 docs refresh template. Corrects several factual errors against the current code, fills in missing detail, and adds a screenshot of the rollback overflow menu. Notable corrections: - Scheduled tasks do not run atomically; only stack editor Deploy and Update, App Store installs, webhook triggers, and image auto-updates pass the atomic flag through to ComposeService. - Rollback lives in the stack editor's More actions overflow menu, not on the action bar directly. The backup timestamp renders as a sub-line of the menu item. - Health probe is a 3-second window with an exit-code check on every container labelled with the compose project name; describe this exactly rather than as 'waits briefly'. - Document where backups live (DATA_DIR/backups/<stack>/), why they are kept outside the compose folder, and that the slot is one per stack with overwrite semantics. - Document the four streamed log markers users see in the deploy progress modal during the atomic flow. - Add a troubleshooting accordion group covering missing menu entry, late crashes outside the probe window, manual-intervention message, and the single-slot retention edge case. * docs(deploy-enforcement): refresh page for v1 and align with current enforcement paths Update the page to match the current pre-flight gate behavior, the v1 modal chrome on the block dialog, and the AccordionGroup troubleshooting pattern used across the v1 docs. Drift items corrected: - Replace the broken vulnerability-scanning/deploy-blocked-dialog.png reference with three fresh captures under docs/images/deploy-enforcement/ (policy list, policy editor, block dialog). - Drop "Recreate from the stack actions menu" and the git-source apply pre-flight claim; neither path runs the gate. - Add bulk label deploy and the auto-update scheduler to the enforced code paths, with a dedicated subsection for the auto-update interaction (alert-and-skip, not 409). - Drop the false claim that severity chips in the block dialog are clickable; the dialog is informational. - Document the compose-parse-fails-closed branch with its synthetic violation label. - Refresh dialog copy to reflect the v1 ModalDestructiveHeader (kicker, title, button variants). - Convert the troubleshooting Q&A into AccordionGroup blocks and add accordions for the compose-parse-error case and the auto-update-skipped case. - Quote the verbatim audit-log summary format. * docs(blueprints): refresh against v1 UI and add federation/state-review coverage * docs(git-sources): refresh page against v1 UI and current behavior Rewrites the page against the v1 docs refresh template (Note tier-gate, sectioned anatomy, AccordionGroup troubleshooting), aligning prose with the live UI labels and the current code paths. Corrections: - Authentication toggle reads "Public (no auth)" / "Personal Access Token" (not "None"), and apply mode "Auto-write files" (not "Auto-write"). - Diff dialog kicker is GIT . PULL PREVIEW; local-edits state opens an Overwrite local edits? confirmation modal whose primary button is Overwrite and apply. - Sidebar pending indicator is a small GitBranch icon, not a brand-color dot, and the image-update dot takes priority over it on the same row. - Pending update banner appears in the panel; Review re-fetches the commit and opens the diff (no client-side payload caching). Adds coverage for: - Anatomy of the panel (pending banner, form, last-applied stat strip, footer actions). - 10-second webhook debounce window. - Pending compose/env content is encrypted at rest in the database, not just the token. - Auth/host failures map to HTTP 400, never 401, so they do not sign the user out. - Per-stack lock serializes pull, apply, and create-from-git so a webhook firing during a manual apply waits rather than racing. - Compose validation has a 10-second budget; clone fetches have a 30-second timeout. - New troubleshooting accordion for Pending commit has changed since this pull was fetched. Recaptures all five screenshots from the v0.74.x production node, signed in as admin: panel, create-from-git tab, pull-preview diff dialog, sidebar GitBranch pending icon, webhook Action select with Git source sync highlighted. * docs(stack-labels): refresh page for v1 sidebar grouping and fleet-action surface - Lead with the v1 behavior the previous page did not cover: the sidebar groups stacks under collapsible label headers (PINNED first, label buckets sorted by stack count desc then name asc, UNLABELED last) with a count chip per group. Trailing colored dots on each row (max 3 + N overflow, paid-only) supplement the headers. - Drop the stale claim that a label-pill filter bar lives between search and the stack list; that UI no longer exists. - Drop the right-click-on-pill bulk actions table (Deploy all / Stop all / Restart all). The legacy per-node action endpoint stays in the backend but no longer has a UI binding, so the page documents only what users can click today. - Document the two Skipper+ Fleet Action cards: Stop fleet by label (name match across nodes, autocomplete, per-node breakdown, HTTP 429 on per-node concurrency) and Bulk label assign (per-node, replace semantics, clear on empty selection). - Document the inline 'New label' form inside the stack right-click / three-dot Labels submenu, the Settings - Advanced - Labels masthead N/50 stat, the LABELS - NEW / EDIT modal kickers, and the LABELS - DELETE - IRREVERSIBLE confirmation copy verbatim. - Document the Fleet Overview Tags multi-select filter (filters by stack labels aggregated across nodes), with cross-link to fleet-view. - Capture every screenshot fresh from production signed in as admin: sidebar-grouping, context-menu-labels, inline-create-form, settings-labels, create-label-dialog, fleet-tags-filter, fleet-actions. Drop the now-stale sidebar-with-labels, sidebar-filtered, and bulk-actions-menu captures. * docs(dashboard): refresh page for v1 layout (status masthead, gauges, fleet heartbeat, restart map) Aligns docs/features/dashboard.mdx with the redesigned Home tab. Replaces the obsolete Recent Activity feed coverage with the actual DashboardActivityCard split (Fleet Heartbeat when remote nodes are registered, Stack Restarts (7d) otherwise) and recaptures every screenshot from the v0.74.x production node. * docs(global-search): refresh page for v1 palette - Note tier and role gating on the Pages list (Auto-Update, Console, Schedules, Audit) so the prose matches what the top bar exposes. - Document the ACTIVE chip on the currently active node row. - Document the 50-result cap counter and the Searching... loading state. - Mention the ~250 ms debounce and clarify that filename matching includes the file extension. - Replace stack screenshot with a redesigned capture and add empty-state Pages and Nodes captures showing the ACTIVE chip. * docs(global-observability): refresh page for v1 layout (masthead, signal rail, filter strip, paused-resume chip) Full rewrite against the current Logs tab and the v1 docs refresh template (hero Frame, sectioned anatomy, AccordionGroup troubleshooting, refresh-cadence table). Replaces the single overview screenshot with seven captures under docs/images/global-observability/ (overview, masthead, signal-rail, filter-strip, feed-bands, paused-resume-chip, error-only-filter), all from the v0.75.x production node signed in as admin with PII scrubbed (profile chip patched to AD, in-feed LAN IPs and third-party hostnames substituted via DOM injection while the stream was paused). Aligns prose with the actual UI labels and code: - Masthead kicker reads LIVE LOGS · NODE · <NAME> with LOCAL for the local node; state word toggles Streaming / Idle / Offline; SESSION uses uppercase letter suffixes (1H 43M / 0M 12S) per formatUptime. - Signal rail tile counts are scoped to the 2000-entry buffer and reset with Clear; CONTAINERS is buffer-bound, not a monotonic accumulator. - Filter strip controls quoted verbatim (Stacks · All / Stacks · n, segmented controls All / Out / Err and All / Info / Warn / Error). - Feed row anatomy: severity dot, timestamp, brand-cyan container name with stack/container tooltip, message tinted by source. Row tint follows detected level, which is regex-based, so an STDOUT line containing ERROR: still classifies as ERROR. - Day bands: NOW, Nm AGO, Nh AGO, calendar date. - Empty states: two-tier kicker over caption (Awaiting events / No matches). - Pause keeps the SSE buffer filling up to the 2000-entry cap; resume pill reads <n> NEW · RESUME and counts the queue, not total arrivals during the pause. - Download filename and row format quoted: sencho-logs-<ISO8601>.txt and [<ISO>] [<stack>/<container>] <LEVEL>: <message>. Documents behavior the previous page never covered: - Active-node scoping; node switch resets the stream and the buffer. - SSE primary transport with 30-second server heartbeat and a 5-second polling fallback against /api/logs/global (server-capped at 500 lines per snapshot). - Initial replay of the last 500 lines per container when the SSE connection opens, so the feed has context immediately. - Display limits (2000 client buffer, 300 rendered rows, Showing last 300 of N overflow notice). - Refresh cadence table covering UI tick, flush cadence, polling cadence, SSE heartbeat, sparkline window, and the Idle threshold. Adds a seven-accordion troubleshooting block (Offline state, gray Idle dot, ERROR-without-tint, growing Resume pill, Clear-cutoff lag, node-switch buffer drop, fleet-wide aggregation expectations). Tightens the closing Note so it makes clear that Notification Log Retention does not govern this live container stream. * docs(alerts-notifications): refresh page for v1 and absorb notification-routing Full v1 template rewrite of /features/alerts-notifications. Bundles in the entire Notification Routing page so a reader sees channels, routing, per-stack rules, and retention in one place; deletes the standalone notification-routing.mdx and points all five cross-link sites at the new in-page anchor. * docs(alerts-notifications): drop "What's not in scope" section The page should describe what Sencho does, not enumerate what it does not ship. Users find missing integrations through the Webhook section and the routing matcher reference; the explicit disclaimer added noise without adding guidance. * docs(audit-log): refresh page for v1 layout, expanded action list, troubleshooting accordion - Clarify that the search/method/date filter strip lives in Table view only. Stream view always shows the unfiltered chronological feed. - Fold the total-entries readout into the card subtitle wording where it actually renders, instead of describing it as a separate header element. - Sharpen the Peak hour off-hours window to the literal 08:00 to 17:59 working window the tile keys off, plus the 5% / 20% failure-rate tints. - Note that the Actors tile names a sample actor alongside the new-IP count. - Expand the example actions list to cover surfaces that have shipped since the last edit: per-service stack lifecycle, node cordon/uncordon, fleet replica role changes, Sencho Cloud Backup operations, Fleet Secrets, and blueprint federation pin updates. - Correct the Settings path: Settings · Developer · Data retention card, Audit log input, Save settings button. - Add a Troubleshooting AccordionGroup matching the rest of the v1-refresh pages: missing tab, filter scope, anomaly thresholds, export cap, and retention pruning. - Replace all four screenshots with fresh captures of the current UI. * docs(multi-node): refresh page for v1 layout, pilot agent mode, refreshed table columns Rewrites docs/features/multi-node.mdx against the current product. The previous page predated the v1 Settings hub redesign and the Pilot Agent enrollment model, so it documented only the Distributed API Proxy add-node flow and missed the new Mode, Endpoint, and Labels columns on the Nodes table. Restructures the page into 13 sections: intro, How it works, the local node, Choose a remote mode (decision table comparing Pilot Agent vs Distributed API Proxy), Add a remote node: Pilot Agent (three steps plus re-enrollment), Add a remote node: Distributed API Proxy (three steps), Switching between nodes, the Nodes table (full column reference), What Settings apply per node (verified against settings/registry.ts), License enforcement across nodes, Editing and deleting nodes, Security (token security, transport encryption, why no application-layer TLS), and Troubleshooting (AccordionGroup matching the v1 template used on audit-log, atomic-deployments, and deploy-progress pages). Refreshes seven screenshots against the production node signed in as admin, scrubbing IPs and usernames before capture: full Nodes panel overview, Generate Node Token card with a placeholder token, Add node modal in Pilot Agent mode, Add node modal in Distributed API Proxy mode (with the inline plain-HTTP warning visible), Edit modal showing the Regenerate enrollment token card for a pilot agent, Pilot enrollment modal with the docker run command, refreshed node switcher popover, and a close-up of the table columns. Drops the obsolete add-node-form.png, http-warning.png, and per-node-scheduling/ folder. * docs(fleet-view): refresh page for v1 layout, expanded tabs, cordon, sheet-based updates - Aligns the Overview, Status, and Node Updates content with today's UI: the masthead's `The fleet` headline plus CPU / MEM / CONTAINERS stat tiles, the eight-tab strip (Overview, Snapshots, Status, Deployments, Traffic, Federation, Fleet Actions, Secrets) with per-tier visibility, and the Check Updates surface that is now a system sheet rather than a modal. - Documents the toolbar (search, sort, filter popover with Status / Type / Severity / Tags sections) and the Grid / Topology segmented control including the topology graph's status pill (Online / Critical / Offline), connector colouring, ReactFlow controls and minimap. - Documents the per-card surfaces that were missing from the prior page: Cordoned badge with cross-reference to Fleet Federation, fleet stack label dots in the drill-down, container drill-down rows (state dot, badge, image, status, open-in-editor hover button), and the Admiral three-dot Node actions menu for cordon / uncordon. - Documents the Node Updates sheet anatomy (Recheck and Update all (n) header actions, four summary cards, node table columns, Update flow, reconnecting overlay timing, admin enforcement) and the GitHub Releases with Docker Hub fallback resolution path with its 30-minute cache. - Replaces every stale screenshot with a fresh capture (overview, topology, drill-down, status tab, node updates sheet) and removes the obsolete files plus the empty docs/images/fleet/ folder. - Reformats troubleshooting as an AccordionGroup matching dashboard, multi-node, and audit-log refreshes. * docs(fleet-backups): refresh page for redesigned fleet and settings UI Replace all six screenshots with current production captures. Update content to reflect the new fleet header card, eight-tab layout, full- page Cloud Backup settings with header stats, and corrected navigation paths. Add cloud backup rows to the access control table. * docs(fleet-backups): convert troubleshooting to AccordionGroup pattern Match the foldable-accordion pattern used across the v1 docs refresh batch. Merges the standalone Cloud Backup troubleshooting subsection into a single Troubleshooting section at the bottom of the page with seven accordions covering skipped nodes, two restore failure modes, three cloud-upload failure modes, and a diagnostic logging entry. * docs(remote-updates): refresh page for v1 sheet, accordion troubleshooting, factual fixes Rewrites the page against the v1 docs refresh template (Note tier gate, sectioned mechanism deep-dive, Frame screenshots with detailed alt text, inline AccordionGroup troubleshooting), bringing it in line with the recently-refreshed fleet-view, fleet-backups, dashboard, and audit-log pages. The page is repositioned as the mechanism deep-dive (prerequisites, what runs on a node during an update, completion and failure detection, recovery actions). The full UI tour for the Node updates sheet remains in fleet-view so the two pages stop overlapping; remote-updates now links into fleet-view#node-updates instead of restating the table anatomy. Captures three screenshots from the production node, signed in as admin: fleet-node-updates.png shows the Node updates sheet with eight nodes and seven remote updates available; local-update-confirm.png shows the LOCAL · UPDATE alert dialog with the Cancel and Update & restart buttons; node-card-update-available.png shows the Opsix card with the Update available pill and the Update to v0.76.7 outline button. Corrects several factual claims that no longer matched the current code: - The remote early-fail threshold is about 3 minutes, matching EARLY_FAIL_MS in backend/src/routes/fleet.ts, not 90 seconds. - The Recheck button sits in the sheet header, not the footer. - The component is a SystemSheet, so the page now consistently calls it the Node updates sheet instead of a dialog, with lowercase "Node updates" and lowercase "Update all (n)" matching the live UI. - Reconnecting overlay polls /api/health every 3 seconds, not "every few seconds". - The local Failed badge surfaces as soon as the helper writes its error file, by the 3-minute mark at the latest. Documents the LocalUpdateConfirmDialog kicker, title, body, and CTA verbatim, the Triggering... loading state on the Update buttons, the four completion signals the gateway accepts (version change, process startedAt change, offline-then-online transition, version at or above the comparison target after 15 seconds), and the 60-second auto-clear of the Updated badge. Drops references to two screenshots that never existed (fleet-node-updating.png, fleet-node-failed.png); the in-flight and failed states are described in prose instead, the same way fleet-view handles them. * docs(scheduled-operations): refresh page for v1 timeline, fleet-wide update action, sheet-based run history Rewrites the Scheduled Operations page against the v1 template (Note tier gate, sectioned anatomy, Frame screenshots, AccordionGroup troubleshooting) applied to sibling pages in this batch. Captures seven fresh screenshots against the production node signed in as admin (timeline, all-tasks, action-picker, create-restart, create-prune, create-scan, run-history) and removes every legacy PNG. Documents the new "Auto-update All Stacks" action that was absent from the page, extends the Skipper allow-list to all four Skipper actions (Auto-update Stack, Auto-update All Stacks, Fleet Snapshot, Vulnerability Scan) and clarifies that the action picker hides operations the active tier cannot run. Corrects several factual claims that no longer matched the code: - Scheduled scan completion is `info`/`scan_finding` on a clean run and `warning`/`scan_finding` when findings are present (not `info`/`system` as previously stated). Cross-link now points at `alerts-notifications#vulnerability-scanning`. - Lifecycle actions (auto_backup, auto_stop, auto_down, auto_start) execute against the local Sencho instance only; only Auto-update Stack / All Stacks have a remote-proxy code path. The page reinstates the guidance to schedule remote lifecycle operations from that node's own UI. - Run history lives in a right-side sheet with a "Schedules › <task> › Runs" breadcrumb and a Download CSV secondary action. - Timeline masthead is described in terms of the v1 visual (`NEXT 24 HOURS` kicker, italic display heading, monospace date range, right-anchored Next pill with countdown, glowing cyan now rail, six-tick bottom axis). * docs(rbac): refresh RBAC & user management page against v1 template Bring /features/rbac onto the v1 docs refresh template (Note tier gate, sectioned anatomy, Frame screenshots, AccordionGroup troubleshooting). Recapture five screenshots from the production node signed in as admin and remove the three stale captures under docs/images/rbac/. Corrections vs. the prior page: - Deployer no longer claims node:read in the permission matrix; the backend grants only stack:read and stack:deploy. - Add the system:registries row (container registry management). - Document the form as inline below the Add user button (not a modal). - Note the (you) marker on the signed-in admin's row and the disabled delete icon on that row. Additions: - Settings nav location and hub-only visibility. - 2FA reset row action with verbatim modal kicker, title, and body. - Five-failure / 15-minute MFA lockout behavior and admin reset recovery. - Token-version session-security table covering deletion, role change, password change, and admin 2FA reset. - SSO password-fields-hidden line quoted verbatim and the per-provider Require MFA toggle. - Audit-log emissions list for every user-management mutation. - API tokens cross-link explaining the user-vs-machine boundary. - Scoped permissions section retightened: scoped role picker is Deployer / Node Admin / Admin only; resource type is Stack or Node. AccordionGroup with eight troubleshooting entries covering missing nav, greyed role options, seat-limit errors, unexpected sign-outs, scoped deployer mismatches, missing shield icon, re-locking MFA accounts, and SSO role drift at provisioning. * docs(2fa): refresh two-factor authentication and admin guide against v1 template Bring /features/two-factor-authentication and /operations/two-factor-admin onto the v1 docs refresh template (Note tier gate, sectioned anatomy, Frame screenshots with descriptive alt text, AccordionGroup troubleshooting, verbatim modal copy with kicker callouts). Recapture every screenshot under docs/images/two-factor-auth/ from a fresh session and add six new captures for surfaces the prior page did not document. Corrections vs the prior pages: - Panel rename: Settings -> Account & Security is now Settings -> Account, under the Identity group of the settings sidebar. Replaced every occurrence on both pages. - Enrol dialog titles match the current modal: Pair your authenticator, Confirm the pairing, Save your recovery codes (was: Set up 2FA, Confirm, Save your backup codes). Step rail 01 PAIR / 02 CONFIRM / 03 ARCHIVE documented. - Manual-entry affordance is the always-visible Secret manual entry row with a copy icon, not the toggleable Can't scan Show secret key link. - Confirm step auto-submits on the sixth digit; no submit button. Verified in MfaChallenge.tsx and MfaEnrollDialog.tsx and called out explicitly. - Authenticator-app list trimmed to match in-app copy (1Password, Bitwarden, Google Authenticator, or any TOTP app). Authy and Microsoft Authenticator dropped because the dialog does not mention them. - Disable dialog: kicker SECURITY MFA DISABLE, title Turn off two-factor, destructive header, Disable button. Replaces the prior Disable 2FA paragraph that did not describe the dialog chrome. - Regenerate dialog: two-step flow with kicker SECURITY BACKUP CODES, Confirm identity then New recovery codes, with the verbatim PREVIOUS CODES HAVE BEEN INVALIDATED warn rail on the show step. Documented that the dialog only accepts a TOTP, not a backup code. - Per-user SSO toggle label corrected: Require 2FA on SSO sign-in (was: Require 2FA even when signing in via SSO). Added the per-provider vs per-user distinction on both pages (admins can also enable Require MFA on the SSO provider config, which is independent of the per-user toggle). - Admin reset modal: verbatim USERS RESET 2FA kicker, Reset 2FA for <username> title, full-body copy reproduced. Documented that the reset bumps the target's token version and invalidates active sessions. Additions: - Sign-in throttle: five failed verifications lock the account for 15 minutes, server returns 423 with Retry-After, UI shows the Retry in MM:SS countdown plus Rate limited label. Lockout recovery section explains that the counter only clears on a successful sign-in, so retries after the window expires re-lock immediately. - Account panel anatomy section enumerates the three rows (Authenticator app, Backup codes, Require 2FA on SSO sign-in) plus the destructive Disable 2FA link, and the masthead 2FA on / BACKUP N left chips. - Recovery codes section now covers all three count states (3 plus, 1 to 2, 0) with verbatim helper text, tone, and the standalone No backup codes left callout that renders at zero. New screenshots for the 2-remaining and 0-remaining states. - Cross-references to the admin operations page (CLI fallback, token version rotation, what a reset changes in the DB), the SSO page, and the RBAC page (per-provider Require MFA toggle, SSO auto-provisioning). Troubleshooting on the feature page rewritten as an AccordionGroup with nine entries: clock drift, wrong account selected, QR will not scan, lost phone with no codes, lost codes with authenticator, ran out of codes, unexpected SSO prompt (with both toggle causes), repeated lockout after the window expires, missing shield icon on Users panel. The admin operations page also gains the SSO + 2FA two-toggles table so administrators can answer the per-user vs per-provider question without context-switching between pages. Six new images added; six existing images replaced. Total 14 captures. * docs(rbac,host-console): drop enforcement-boundary detail from tier-gate notes Operator-facing docs should state tier or role requirements once, in plain customer-facing language, and leave the enforcement chain to the source. Two surfaces on the v1-refreshed pages over-specified the gate: - `features/rbac.mdx::Scoped permissions`: the Note enumerated both the UI hide on Skipper and the `/api/users/:id/roles` write rejection. The first half ("Scoped permissions require Admiral.") is the operator-relevant fact; the rest reads as a fence specification, which is awkward for an open-core product where the gate is readable in source anyway. Trimmed to just the tier claim. - `features/host-console.mdx::Availability`: the paragraph already says who can use the console and that the Console tab is hidden on Community or Skipper. The trailing "Attempting to access the console endpoint directly without the correct license or role is rejected" is the same bypass-prevention coda. Dropped. No functional behavior change; the gates themselves are untouched. * docs(sso): refresh SSO & LDAP authentication page against v1 template Rewrites docs/features/sso.mdx against the v1 docs refresh template (intro + tier callout, sectioned Configuration anatomy, Frame screenshots, AccordionGroup troubleshooting), bringing it in line with the previously refreshed two-factor-authentication and rbac pages on this branch. Recaptures all four screenshots from the production node signed in as admin: sso-settings (overview with the five collapsible provider cards), sso-settings-ldap (LDAP form expanded), sso-settings-oidc (Google form expanded), sso-settings-custom-oidc (Custom OIDC form expanded with all eleven fields). Refreshes the Settings UI section to match the redesigned panel: each provider is a collapsible card with an Active badge on the header, an enable / disable toggle pill, and a footer with Save, Test Connection (green check or red X next to the button), and Remove (only after a config has been saved). Documents the static callback-URL helper that sits below all five cards. Clarifies that the per-OIDC claim mapping environment variables (SSO_OIDC_*_ID_CLAIM, *_USERNAME_CLAIM, *_EMAIL_CLAIM) are accepted for Google, GitHub, and Okta, not just Custom OIDC. The Settings UI hides those fields on the presets because the defaults match. Converts the troubleshooting section to an AccordionGroup with five entries (Test Connection discovery failure, issuer validation error, wrong username or missing email after sign-in, invalid redirect URI, SSO buttons missing on the login page). Cross-links the operations troubleshooting page for setup-time errors. Tightens the LDAP TLS env var note to spell out the literal string 'false' requirement. Syncs the Combining SSO with 2FA section to use the live toggle label 'Require 2FA on SSO sign-in'. * docs(sso): drop the Community-tier Custom OIDC workaround tip The Tip walked through how a Community-tier operator could integrate Google, GitHub, or Okta by pointing Custom OIDC at the provider's discovery URL, bypassing the Skipper preset gate. Operator docs should state the tier rule once and stop; they should not describe how to circumvent it. The tier matrix above the removed block already names which providers are paid; the Custom OIDC row already lists "any spec-compliant OIDC provider" as its scope. That is enough. * docs(vulnerability-scanning): refresh page for v1 UI and corrected tier mapping The page was last revised before the v1 visual redesign and before the tier-mapping changes shipped in v0.81.2 (open Community access to secret scanning, compose misconfig scanning, scan history, and scan comparison). This refresh: - Rewrites the tier matrix to match the shipped Community / Skipper / Admiral split. Secret detection, compose misconfig scanning, scan history, scan comparison, and misconfig acknowledgements are now correctly marked as Community. Scheduled fleet scans, scan policies with block_on_deploy, SBOM, SARIF, and Trivy auto-update stay paid. - Drops two stale Notes that said secret detection and compose misconfig scanning required Skipper or Admiral. The page now states each tier requirement once, in plain language. - Refreshes all six existing screenshots from the production node: resources-badges, scan-details-sheet, scan-history-sheet, scan-compare-sheet, security-settings, app-store-toggle. - Adds a new scan-config-button screenshot showing the stack-page overflow menu where Scan config now lives. - Describes the scan drawer header accurately: Re-scan + Compare + CSV + SARIF as top-level buttons, with SBOM as a separate button below the summary. - Updates the compose misconfig flow to point at the stack overflow menu (not the Deploy controls). - Converts the troubleshooting section to a single AccordionGroup per the v1 template, and audits each entry for legacy phrasing and the removed tier claims. - Adds a TRIVY_BIN reference to the How it works section so operators know about the host-binary override. * docs(cve-suppressions): refresh page for v1 UI and corrected suppression specifics - Recapture all three screenshots from the production node signed in as admin under `docs/images/cve-suppressions/` (`settings-panel`, `create-dialog`, `suppressed-row`). The previous file referenced three image paths that did not exist in the repo. - Align prose with the actual UI labels: - Dialog kicker `SUPPRESSIONS . NEW`, title `New suppression`. - Field labels match the form: `CVE or advisory ID`, `Package (optional)`, `Image pattern (optional)`, `Reason`, `Expires in (days, optional)`. - Remove confirmation reads `Remove suppression` with kicker `SUPPRESSIONS . REMOVE . IRREVERSIBLE`. - Factual corrections: - Fleet sync truncation cap is 5,000 rows (not 10,000). - State the admin-role requirement once in the lead Note. - Drop references to a `Fleet . Sync status` page and a `Reanchor` button; neither exists in the UI. The reanchor flow is an admin API call and is documented in /features/fleet-sync. - Sharpen the specificity scoring section (package + image scores 3, package only 2, image only 1, neither 0) so the order matches the read-time filter logic. - Note that the image-pattern glob is case-sensitive. - New coverage: - Suppressing directly from a scan result, including which fields are read-only in that inline flow and when to fall back to Settings to broaden scope. - The `replicated` and `expired` row badges in the panel. - Hovering the package column on a suppressed row to surface the Reason. - Two distinct read-only modes: viewing a remote node from the hub (panel hidden, banner shown) versus signing into a replica instance (panel visible, read-only). - SARIF export carries suppressions through as `kind: external, status: accepted`, cross-linked to the Vulnerability Scanning page. - Convert troubleshooting to AccordionGroup with six entries; update the truncation entry to reflect the 5,000-row cap. * docs(private-registries): refresh page for v1 UI and fleet-wide credential model Rewrites the page against the v1 docs template (Note tier gate, opening Frame, sectioned anatomy, AccordionGroup troubleshooting) and replaces every screenshot with a fresh capture taken against the current product. Corrects several factual claims that no longer matched the current code: - Registries are stored once on the control instance and applied fleet-wide, not configured per node. The old Multi-node behavior section and the matching troubleshooting entry described a per-node model that the product no longer has. - The Registries section is hidden on remote nodes (global scope) and on Sencho versions that do not surface the feature. New troubleshooting entries explain both visibility states. - The feature is admin-only on Admiral. Non-admin operators do not see the section even on Admiral; previous copy implied any Admiral license user could manage credentials. - Registry endpoints are not reachable from API tokens; only an admin browser session can manage credentials. The Security section now states this without naming internal route paths. Documents UI behavior the previous page omitted: the inline form (not modal), the four type-specific form variants, the Docker Hub read-only URL field, the destructive delete confirmation with its stack-pull warning, the masthead REGISTRIES count, and the empty-state callout copy. Screenshots replaced: - registries-overview.png: section with one configured GHCR card and the masthead stat at one. - registries-empty.png: empty state with the Add registry button and callout. - registries-add-form.png: inline form with the Docker Hub default and the read-only URL field. - registries-ecr-form.png: form switched to ECR, showing the AWS Region field and the relabelled AWS credential inputs. - registries-card-detail.png: card close-up with the three action icons and the metadata row. - registries-delete-confirm.png: destructive ConfirmModal with the kicker, title, and stack-pull warning body. - registries-with-entry.png removed (superseded by registries-overview.png and registries-card-detail.png). * docs(auto-update): refresh readiness page for v1 redesign Bring the Auto-Update Readiness doc in line with the shipped UI: - Replace the hero screenshot with a fresh capture of the redesigned board (italic-display hero, brand-cyan accent, per-node groups with local/remote pills, dashed-border changelog separator). - Rewrite the card-anatomy list. Drop the rollback-target bullet (the field exists in the backend payload but is not rendered). Add the "Rebuild available" inline label and the primary-image / multi-service count line. - Rewrite the risk-tags table as a risk-badges table using the actual badge labels and colors emitted by the UI (Safe / Review / Blocked with the corresponding icons; Digest rebuild for non-semver tags). - Add an Empty state section and document the per-node group header. - Tighten the hero subtitle paragraph to match the actual UI string (only major-bump count is surfaced separately; preview failures are not). - Fix workflow step 4: major-bump apply path is the stack lifecycle Update action, not the Schedules editor (a scheduled task hits the same block). - Add the 2-minute manual-refresh cooldown to the Recheck workflow. - Remove the broken cross-link to the non-existent /features/image-update-detection page and inline the 6-hour cadence fact from ImageUpdateService.INTERVAL_MS. - Convert troubleshooting to AccordionGroup format per the troubleshoot ing convention used on /features/deploy-progress. - Sync the Auto-Update entry in /features/overview.mdx to the new badge labels and the corrected hero-counter description. * docs(auto-update): fix Auto-Update entry point in Workflow step 1 Workflow step 1 said "Open the Auto-Update view from the sidebar." The Auto-Update view is opened from the top nav strip (alongside Home, Fleet, Resources, App Store, Logs, Schedules, Console, Audit). The sidebar carries the stack list and the per-stack right-click / kebab context menu that toggles auto-updates on or off; it does not house the Auto-Update top-level view. * docs(auto-update): trim enforcement detail from per-stack control note State the tier requirement once and stop, per Directive 27. The "The toggle does not appear on Community" sentence enumerates the enforcement effect of the gate, which the source already reflects; operator docs do not need to narrate it. * docs(auto-heal): refresh page for v1 UI and policy hardening Rewrite Auto-Heal Policies docs against the current Stack Monitor sheet: corrects the Max restarts / hr field label, documents the per-policy enable toggle, the consecutive-failures pill, the full Recent activity action set (including Docker unavailable), the 30s evaluation cadence, multi-node behavior, notification dispatches, and the dashboard Configuration status counter. Replaces the broken /images/auto-heal-policies/policy-sheet.png reference with three fresh screenshots captured against a live node: the sheet on the Auto-heal tab, a single policy row, and the expanded Recent activity panel. * docs(webhooks): refresh page for v1 UI, correct tier and add Git source sync - Fix tier note: gate is Skipper or Admiral, management is admin-only. - Update Settings path to Settings -> Alerts -> Webhooks; document the read-only Node field and the green secret-reveal callout. - Add the missing Git source sync action and the git-pull override value. - Refresh the configured-webhooks card description: action/stack/node badges, On/Off toggle, copy URL, and the Recent executions disclosure. - Tighten the trigger section with a constant-time signature check note and a status/body/meaning response table. - Add an Accordion troubleshooting block covering common signature failures, the 404 case, no-op actions on 202, and git-pull prereqs. - Re-capture all three screenshots from the v1 UI. * docs(webhooks): wrap troubleshooting accordions in AccordionGroup * docs(sidebar): refresh page for v1 redesign with filter chips, bulk mode, row anatomy, and troubleshooting Rewrites the Stack Sidebar page against the live v1 sidebar and the v1 docs refresh template (Frame screenshots, Note tier callouts, AccordionGroup troubleshooting). Recaptures all four existing screenshots and adds three new captures: filter chips, row anatomy, and bulk mode. Adds coverage for features the previous page omitted entirely: the ALL / UP / DOWN / UPDATES filter chips with their counts cap and collapse toggle; bulk mode (B key, sticky toolbar with Start / Stop / Restart, and Update gated on Skipper or Admiral); stack-row anatomy (status pill, label dots with +N overflow, image-update dot vs Git source icon priority, hover kebab); the Auto-update toggle, Schedule task, and Open App entries in the context menu; the B shortcut for bulk mode. Corrects three claims that no longer matched the code or UI: Auto-Heal is gated on Skipper or Admiral, not universal; the global Ctrl+K opens the command palette, not the sidebar search box; the activity footer kicker reads LIVE / IDLE with the verbatim copy from SidebarActivityTicker. Documents the in-menu ↗ and L › glyphs as visual hints rather than global keybindings to match useStackKeyboardShortcuts.ts. * docs(sidebar): trim enforcement-effect sentence from context-menu tier note State the Skipper / Admiral requirement once and stop, per Directive 27. The "They do not appear in the menu on Community" clause described the enforcement effect alongside the gate, which the directive bans in operator-facing docs. * docs(host-console): refresh page for v1 UI and clarify shell metadata Rewrite the Host Console page to match the current Cockpit layout (masthead + terminal well + chip strip), replace the legacy PowerShell screenshot with a fresh bash capture, and document the masthead tone states, kicker, metadata pills, and session/heartbeat behavior. Trim the security section to state the tier and role rule once. * docs(licensing): refresh page for v1 UI, corrected pricing, and trial flow Rewrites the Licensing & Billing page to match the redesigned v1 Settings layout. The previous draft still described the legacy Settings Hub: in-app "Upgrade your plan" Skipper/Admiral cards, "Start monthly trial" / "Start annual trial" buttons, the "Have a license key?" field, "Manage Subscription" with a capital S, "Deactivate License" as the button label, and the license-active.png asset rendering the literal "Sencho Pro" string in the card title. None of that exists in the current product. - Refreshes the Plans table to the live pricing on sencho.io/pricing and adds an Enterprise mention with the floor price ($3,500/year). Skipper now $11.99 annual / $14.99 monthly / $449 lifetime, Admiral now $69.99 annual / $89.99 monthly / $2,499 lifetime. - Rebuilds the Feature breakdown from a code-level audit of every requirePaid, requireAdmiral, requireScheduledTaskTier, and requireTierForSsoProvider call site in backend/src/routes, not from the marketing page. Notable code-grounded items: CVE suppressions on Community (no requirePaid guard), manual fleet snapshots on Community (scheduled snapshots on Skipper), Sencho Mesh under Admiral (entire mesh.ts router is requireAdmiral), and scheduled-task tiering names update/scan/snapshot as the Skipper subset with everything else under Admiral. - Rewrites the Free trial flow end to end. The previous steps told operators to click in-app "Start monthly trial" or "Start annual trial" buttons; no such buttons exist. The new flow starts on sencho.io/pricing, switches to the Annual or Monthly tab, clicks "Start 14-day trial" on the Admiral card, completes the Lemon Squeezy checkout (card-required, no charge before day 14), and pastes the issued key into Settings -> License -> License key. - Adds a new "The Plan section" anatomy block describing the masthead SCOPE / PLAN / DURATION (or RENEWS, TRIAL, STATUS) stat pills and the Plan card fields (Customer, Product, masked License key, status helper). - Adds a new "License states" reference table covering Community / Trial / Active subscription / Active lifetime / Expired / Disabled, what each surface renders, and which of the Plan / Activate / Pricing sections is visible in each state. - Corrects every UI label that drifted: section heading is Activate, field label is License key (not "Have a license key?"), buttons are Manage subscription (lowercase s) and Deactivate (not "Deactivate License"), and the action-row hint reads "Lemon Squeezy manages billing". - Documents the redesigned profile dropdown: identity header with initials chip, role badge, and tier badge, then Settings, conditional Billing, Documentation, Feedback, an Appearance segmented control, and Log Out. Billing only appears when the license is an active non-lifetime subscription. - Replaces all four screenshots under docs/images/licensing/: license-admiral-active.png (production Admiral lifetime view), profile-menu.png (redesigned popover), and two new captures for the Community-tier surfaces (license-activate-section.png, license-community.png). Removes the stale license-active.png (legacy "Sencho Pro" card) and profile-billing.png (legacy dropdown). * docs(settings-reference): refresh page for v1 UI with new sections and masthead Rewrites docs/reference/settings.mdx against the current Settings Hub so a reader encounters an accurate map of every section. Adds the previously missing **Cloud Backup** and **Security** sections, restructures **System Limits** into Host thresholds and Docker hygiene subsections (GiB units, "Global crash capture" toggle), fixes the Account password minimum to 8 chars and documents the two-factor subsection, refreshes License/Routing/Webhooks/App Store with the field labels actually rendered today, and documents the masthead pills (SCOPE/NODE, EDITED, plus the per-section stats like 2FA, PLAN, CHANNELS, ROUTES, WEBHOOKS, LABELS, TRIVY, POLICIES, PROVIDER, USED, SNAPSHOTS, DEV MODE). Replaces five existing screenshots that predated the v1 redesign and adds five new captures: Account with the 2FA card, License panel, System Limits with both subsections, Security with the Trivy installer, and Cloud Backup with Sencho Cloud Backup provisioned. All shots taken against the production node. * docs(licensing): drop billing-provider name from operator-facing copy The previous draft named the third-party billing provider in nine places (checkout, receipt email, error toast verbatim, Customer / Product field descriptions, the action-row hint, the billing portal, and the validation API). Operator docs don't need to advertise which vendor sits behind the checkout, billing portal, and validation calls. Rewrite each instance to describe what the operator sees and does without naming the upstream service. * docs(node-compatibility): refresh page for v1 UI with lock card visuals and current capability list - Replaces the legacy "dim + blur + pill overlay" description with the current CapabilityGate behavior: a centered lock card with an Unplug icon, title "<feature> is not available on this node", and a body line that names the node and its running version. - Corrects the tier-interaction section: on the wrong tier the entry point is hidden entirely, so the lock card only appears for users who already cleared the license gate. - Documents the public /api/meta endpoint, the 5-minute success cache, the 30-second failure cache, and the lazy-fetch behavior visible in the switcher (the version pill appears once a node has been visited). - Refreshes the capability table against the current CapabilityRegistry list, adding container-exec and vulnerability-scanning, with a note that vulnerability-scanning is only advertised when Trivy is installed. - Adds three production screenshots captured on the live fleet: switcher popover with mixed-version pills (one node on v0.76.9, rest on v0.81.11), a real lock card on an older pilot agent, and the Connection Details panel from Settings · Nodes. * docs(security): refresh security architecture page for v1 UI Add Fleet Secrets and Webhook signatures cards plus tier-matrix rows for shipped-but-undocumented features. Rename SSO presets from "one-click" to "preset providers" (presets still require OAuth-app provisioning on the upstream IdP). Update settings paths to the v1 middle-dot convention: Settings · Users, Settings · Account, Settings · Developer · Data retention. Extend the encryption-at-rest list with registry credentials and Fleet Secrets bundle payloads (both sealed with the same AES-256-GCM data key) and clarify the password section with bcrypt cost factor 10. Add a Webhook signature authentication subsection covering the per-webhook HMAC-SHA256 secret, one-shot display, masked preview thereafter, and constant-time comparison on inbound triggers. Replace the API Tokens screenshot with a fresh capture against the v1 Settings · Identity · API Tokens panel. * docs(security-advisories): retire reference page The reference/security-advisories page does not survive the v1 docs refresh: - Misuses the term "Security Advisories", which industry-wide refers to published notices for confirmed product CVEs (ID, severity, affected versions, fix version, remediation). The retired page was a narrative changelog of internal hardening work between v0.19 and v0.25.2. - The narrative is also frozen at v0.25.2 (April 2026) while current release is v0.81.11. Refreshing it would require backfilling ~56 release entries' worth of hardening copy. - The framing is uniformly "improved from prior behavior" (minimum 8 characters up from 6, 1-year token expiry previously without expiry, CORS previously allowed all origins, users should upgrade promptly). Sencho has not shipped publicly; there are no users to address as upgraders. All operationally relevant content already lives elsewhere: the security architecture page covers the current posture, verifying-images covers the supply-chain attestations, cve-suppressions covers operator acknowledgment, vulnerability-scanning covers the in-app scanner, and contact + the security architecture page both surface the disclosure path. Published Sencho-product advisories, when any exist, will appear on the GitHub Security tab, which is already linked from those pages. Inbound-link audit returned a single hit on the nav entry itself; no other doc, README, or operator artifact deep-links the slug. * docs: rewrite Pilot Agent page with deep architecture and operations reference Reframes docs/features/pilot-agent.mdx as the architecture-and-operations companion to the operator walkthrough in Multi-Node Management. Adds a mental model section, an explicit security and trust model, a full agent env-var reference, an honest limitations list, and a 5-item FAQ. Verifies every constant and label against the current backend source. Refreshes four production screenshots (admin login, scrubbed) and resolves the previously-broken /images/pilot-agent/enrollment-dialog.png reference. Adjacent edits keep the cross-linking coherent: - multi-node.mdx adds a one-line forward link to the rewritten page - security.mdx adds a Pilot Agent tunnel credentials subsection * docs(fleet-federation): deep rewrite with production screenshots Rewrites the Fleet Federation page against the v1 docs refresh template following the recent fleet-view, pilot-agent, and multi-node refreshes. Doubles the page length (92 to 204 lines) while keeping the cut-line v1 MVP scope: operator-driven placement controls (cordon + pin) for Blueprints, no expansion into mesh/sync/pilot territory. What changed: - Adds four production-captured screenshots under docs/images/fleet-federation/: the Federation tab with a cordoned node populated, the node-card kebab menu showing the Cordon node entry, the cordon confirmation dialog with a reason filled in, and a node card displaying the Cordoned pill. - Expands the page to eleven sections: opening summary, philosophy (kept), key capabilities, prerequisites, step-by-step usage with embedded screenshots, behaviour and lifecycle table, security and audit, limitations and non-goals (expanded), practical workflows (new: OS patching, host-to-host migration, gateway pinning), troubleshooting (eight accordions, up from five), and a Where Federation fits cross-link table. - Documents the exact production UI strings observed: the cordon dialog description, the uncordon confirmation copy, the reason field cap (256 chars), and the audit log action names (node.cordon, node.uncordon, blueprint.pin). - Documents the audit visibility surface so operators know how to filter the Audit view for cordon and pin history. - Adds eight cross-links to sibling pages (Fleet View, Multi-Node, Pilot Agent, Mesh, Fleet Actions, Fleet Sync, Blueprints, Licensing) with one-line scope contrasts so newcomers can place Federation in the broader fleet picture. - Tightens lifecycle table to operator-relevant terms (no DB column names) and audit section to operator-facing wording (no middleware names), keeping the page operator-focused rather than implementation-focused. Validation: - Captured screenshots against the production node logged in as admin, using Playwright MCP. Cordoned and pinned actions reverted; audit log confirmed the matched cordon/uncordon pair. - Verified every cross-link target exists in the v1-refresh worktree (/features/fleet-view, /features/multi-node, /features/pilot-agent, /features/sencho-mesh, /features/fleet-actions, /features/fleet-sync, /features/blueprint-model, /features/licensing). - Compliance: no em dashes, no PII, no "previously"/"used to" framing, no fence-spec language, tier rule stated once in plain language. * docs(fleet-federation): drop fence-spec phrasing in the open-core context Sencho is open-core: anyone can clone the repo and read the tier gate. Operator docs that name exactly where the UI gate sits ("hidden at the Community and Skipper tiers", "lower-tier users do not see the toggle", "only the Federation tab is gated") work as a dig-target for a tech-savvy reader and undercut the open-core posture. Directive 27 already bans enforcement-chain spelling; the open-core threat model makes the same phrasings risky even when they describe UI surfaces rather than route guards. Removes three instances of the pattern on this page: - Top Note callout: drops "The tab is hidden at the Community and Skipper tiers." Keeps the one-line requirement: "Federation is an Admiral feature. Cordon and pin actions require an admin user role." - Security and audit section: drops the sentence enumerating which UI affordances are hidden from which tiers. Keeps the customer-visible behavior (the Cordoned pill stays visible at every tier as a read-only signal). - Troubleshooting "Federation tab is not visible" accordion: rewrites to lead with the requirement and the role check, drops the "Federation is hidden by design" and "only the toggle and the Federation tab are gated" phrasings. Other claims on the page unchanged; rule is still stated once in plain language at the top of the page. * docs(fleet-sync): deep rewrite with production screenshots Replace fleet-sync.mdx with a verified end-to-end reference. The previous page named two replicated resources but the code syncs three, described a sync-status panel and a fleet-vs-node scope picker that do not exist in the shipped UI, and was missing prerequisites and several edge cases. Highlights of the rewrite: - Names all three replicated resources (scan policies, CVE suppressions, misconfig acknowledgements) and treats them uniformly. - Drops the sync-status-panel and node-scope-picker UI claims; both move to the Limitations section as honest caveats. - Adds prerequisites covering the paid-tier requirement on the control, admin-role requirement, proxy-mode remotes, and reachability. - Expands lifecycle coverage: per-node serialised pushes, add-node backfill, monotonic pushedAt, per-resource watermarks, identity-drift notifications, the 5000-row truncation cap, stale-target warnings, audit-log entries on the replica. - New "Where Fleet Sync fits" closing table cross-linking to Fleet View, Multi-Node Management, Pilot Agent, Vulnerability Scanning, CVE Suppressions, Fleet Federation, Fleet Actions, and Licensing. - Two fresh production screenshots: control Security panel and the "Scanner is per-node" callout shown when proxying to a remote. * docs(fleet-actions): deep rewrite with production screenshots Three cards are documented end to end: Stop fleet by label, Bulk label assign, and Prune Docker resources fleet-wide. Adds the execution-path distinction (control-orchestrated fan-out vs single-node proxy), per-card behaviour and partial-failure semantics, prerequisites, limitations, practical workflows, an Accordion troubleshooting section, and a Where Fleet Actions fits comparison table linking the surrounding Fleet view features. Corrects the prior page's tab-neighborhood claim, confirm-dialog wording, autocomplete-vs-request scope, and missing batch ceiling. Replaces the ten-day-old single screenshot with five fresh production captures under docs/images/fleet-actions/. * docs(fleet-secrets): deep rewrite with production screenshots Full rewrite of /features/fleet-secrets matching the fleet-actions structure. Replaces the sparse v1 page (no Frames, inline Q&A) with a gold-standard layout: opening Frame, single Note for the tier gate, 'What it covers' table, mental model, prerequisites, create + edit + versions + push (Target / Preview / Results) sections each with a production Frame, Import from stack section, behaviour and lifecycle table, audit-trail mapping with the six exact audit strings, limitations and non-goals, practical workflows, AccordionGroup troubleshooting, and a Where-it-fits cross-link table. Adds six fresh production screenshots under docs/images/fleet-secrets/ : overview, create, versions, target, preview, and results. Documents the Import-from-stack flow (depends on the bundle editor's new Import action) and uses the post-rename 'Send' wording on the bundle-row action (depends on the aria-label fix). Corrects three factual drifts vs the code: env-key regex described as 'letter or underscore, then letters/digits/underscores; case- sensitive' to match ^[A-Za-z_][A-Za-z0-9_]*$ ; documents only the 'ok' and 'failed' status pills (the 'skipped' enum value is unused); replaces the bogus 'stack not found' troubleshooting entry with the real 'env file not declared' cause. Drops the fence-spec phrasing 'The tab is hidden on Community.' per Directive 31; the tier requirement is now stated once in plain language. * docs(sencho-mesh): deep rewrite with mental model, lifecycle, security, screenshots Replace the feature-reference page with a deep product + technical guide. Adds: - Opening hook framing audience and problem (cross-node service-to-service without a separate VPN or service-mesh sidecar). - Mental model: three moving parts (sencho_mesh bridge, alias registry, cross-node transport) with direction-of-flow described in prose. - Key capabilities, prerequisites, step-by-step usage with inline screenshots. - Full lifecycle section covering opt-in, opt-out, sticky stack-stopped state, peer reconnect, and the proxy-mode bridge with its real default (persistent, env-override for idle). - Security and trust boundaries split into authentication, inbound exposure, encryption, audit, and app-layer caveats. - Limitations and non-goals: one-alias-per-port, port 1852 reserved, central-relay for remote-to-remote, shared 1024-stream pool with the Pilot tunnel, no L7, host-network unsupported, in-memory activity log. - Three concrete workflow examples and a complete troubleshooting accordion (every data-plane reason, every probe stage, every unreachable cause) plus a Common questions FAQ. - Where Mesh fits CardGroup linking Pilot Agent, Multi-Node, Federation, Licensing. Corrections vs prior text: - Tab is labelled Traffic in the UI (not Routing); all navigation references updated. - Proxy-mode bridge default is no idle close (env-overridable to opt into idle teardown); prior 5-minute-teardown claim removed. - Audit trail scope tightened: only opt-in / opt-out write durable rows; tunnel-state and probe events live in the in-memory activity log. Adds seven production screenshots under docs/images/sencho-mesh covering Table view, opt-in sheet, graph (Tunnels and Aliases edge modes), Diagnostics, activity log, and per-stack topology. * docs(blueprints): add missing detail-state-review screenshot Captures the Blueprint detail sheet with a deployment row in the "Awaiting confirmation" status (stateful first-deploy gate), to fix the broken image referenced at blueprint-model.mdx:132. mint broken-links now reports zero broken references. * docs(blueprints): deep rewrite with mental model, lifecycle, security, prerequisites Restructures the Blueprints page against the v1-refresh template used by the recently-refreshed mesh, secrets, and atomic-deployments pages. Adds a mental model, prerequisites table, lifecycle and status-transition map, security and trust boundaries section, practical workflows, common questions accordion, and a Where Blueprints fits CardGroup. Removes the internal-style rollout and watch-plan section. Replaces all nine production screenshots with fresh captures against the production node signed in as admin, and adds two new captures (federation pin policy table, stateless eviction dialog). Rewrites the tier-gate Note to drop the fence-spec phrasing that violated Directive 31. Every retained claim is anchored to current backend or frontend code. * docs(pilot-agent): recapture enrollment dialog with compose payload Replaces the pre-0.84 docker-run capture with the current dialog (Compose file, two-step instructions, "Copy compose file" button) and refines the alt text to describe the captured content. URL and token redacted to placeholder values during capture.	2026-05-20 08:43:18 -04:00
Anso	282ab8d844	fix(pilot): let SENCHO_PUBLIC_URL override the request Host in enrollment (#1122) ... The enrollment minter inferred SENCHO_PRIMARY_URL from the request Host header, which baked loopback or LAN addresses into the compose YAML when the admin opened Add Node on the central's own machine. Pilots on a different network (a public cloud VPS, for example) cannot dial that. SENCHO_PUBLIC_URL on the primary now wins when set and well-formed (http(s)://, no loopback). Trailing slashes are stripped. Falls back to the request Host when unset or invalid.	2026-05-20 03:35:25 -04:00
Anso	3ad6ea9c5d	feat(pilot): make Docker Compose the canonical pilot enrollment payload (#1121) ... * feat(pilot): make Docker Compose the canonical pilot enrollment payload The Add Node dialog for a pilot-agent now returns a Compose snippet instead of a single-line docker run command, and the enrollment dialog walks the operator through a save-and-up flow. The compose project name and container name align with what SelfUpdateService looks up at boot, so a Compose-deployed pilot can be updated remotely through the Fleet view without intervention on the remote host. Docs (pilot-agent, remote-updates) were rewritten to match. * test(e2e): align pilot enrollment spec with Compose payload The spec was written against the docker-run payload; it now asserts the Compose YAML the dialog renders.	2026-05-20 02:05:21 -04:00
Anso	23bbee4f45	feat(mesh): replace host-mode with shared sencho_mesh Docker network (#1009) ... * feat(mesh): replace host-mode with shared sencho_mesh Docker network Phase D of the mesh redesign: drop the operator's `network_mode: host` requirement and the `host-gateway` extra_hosts pattern that did not work on cloud iptables-restrictive distros (OCI, etc.) or Docker Desktop. Each Sencho creates a shared `sencho_mesh` Docker bridge network on boot (default subnet 172.30.0.0/24, override via SENCHO_MESH_SUBNET), pins itself at `<network>+2`, and attaches every meshed user service to the same bridge. Compose overrides now emit IP-based `extra_hosts` plus a top-level `networks` block declaring `sencho_mesh` external. Override delivery: central renders for local stacks; for remote stacks it sends the fleet alias list to the remote's new `PUT /api/mesh/local- override/:stackName` endpoint, which renders against the remote's OWN local senchoIp and writes under its OWN DATA_DIR. Each node may use a different subnet without coordination beyond the env var. Opt-in / opt-out now trigger an automatic redeploy of the affected stack via the existing deploy code path (local: ComposeService; remote: HTTP POST through proxyFetch). The frontend opt-in sheet shows a confirmation modal (ConfirmModal) before the mutation. Failed redeploys emit both a mesh activity event and a durable audit-log row. Hardening: - Reserve port 1852 at opt-in (prevents user containers from racing the Sencho API listener). - ensureMeshNetwork refuses to continue if `sencho_mesh` exists with a mismatched subnet rather than silently routing to the wrong IP. - Idempotent network connect/disconnect helpers in DockerController. - optInStack rolls back the DB row if the just-inserted stack's override push fails (no half-states surviving across calls). - regenerateOverridesForNode runs in parallel and skips the just- pushed stack on opt-in. Operator template: drop `network_mode: host`, restore `ports: ["1852:1852"]`. Mesh now works identically on Linux LAN, OCI, and Docker Desktop without firewall changes. Docs: rewrite docs/features/sencho-mesh.mdx around the shared bridge network, document SENCHO_MESH_SUBNET, surface the host-network-service opt-in restriction, and cross-link with the Pilot Agent docs. BREAKING CHANGE: the operator's `docker-compose.yml` no longer uses `network_mode: host`. After upgrading, redeploy any meshed stacks once so they pick up the new IP-based override and join `sencho_mesh`. * fix(mesh): wrap stackName with path.basename in local-override fs ops CodeQL flagged js/path-injection on the new applyLocalOverride and removeLocalOverride methods because they are publicly reachable and its data-flow model does not recognize isValidStackName / isPathWithinBase as sanitizers. The validation IS sufficient (the allowlist regex blocks path separators, the path-prefix check blocks escape), but path.basename is a model CodeQL recognizes and is purely defensive: for any input that already passes isValidStackName, basename is the identity.	2026-05-09 00:11:09 -04:00
Anso	3d9489648e	fix(pilot): harden outbound reverse-tunnel against resource exhaustion (#979) ... * fix(pilot): cap tunnel frame size, concurrent streams, and stream idle time Pilot tunnels carry every HTTP, WS, and Mesh-TCP byte for a remote node through a single multiplexed WebSocket. A buggy or compromised peer that sent oversized frames, opened streams in a tight loop, or left streams parked indefinitely could exhaust gateway memory. Three protocol-level limits applied symmetrically on both ends: - MAX_FRAME_SIZE_BYTES (8 MB): set as the ws maxPayload on the gateway-side WebSocketServer and the agent-side WebSocket client, plus a defense-in-depth length check in decodeBinaryFrame and decodeJsonFrame. - MAX_STREAMS_PER_TUNNEL (1024): bridge refuses new loopback HTTP / upgrade / TCP allocations with 503 once at the cap; the agent rejects new incoming http_req / ws_open / tcp_open with the appropriate error frame. - STREAM_IDLE_TIMEOUT_MS (10 min): every stream gets a per-stream timer refreshed on each inbound or outbound activity. Expiry tears down the local half and notifies the peer. The constants are colocated in pilot/protocol.ts so any future agent build picks them up via the protocol module. * fix(pilot): rate-limit pilot enrollment endpoints to 10 per minute Pilot enrollment mints a JWT and writes a pilot_enrollments row, both privileged operations that should not share the global API budget (200/min). Add a dedicated express-rate-limit instance keyed by user or IP at 10/min in production (100/min in dev so the local enrollment test loop is not throttled). The limiter is applied directly on POST /api/nodes/:id/pilot/enroll. On POST /api/nodes the limiter's skip function reads the parsed body and exempts proxy-mode creates; only requests that resolve to mode=pilot_agent count against the enrollment budget. * fix(pilot): cap concurrent pilot tunnels system-wide The PilotTunnelManager held an unbounded Map of bridges. A reconnect storm, runaway enrollment, or operator misconfiguration could grow the map without limit, even though every tunnel still carries a valid JWT. Cap at 256 concurrent tunnels per primary instance: - Soft warning at 128 (logged once per crossing). - Hard refusal at 256: registerTunnel throws PilotTunnelCapacityError and the upgrade handler closes the WebSocket with 1013 (Try Again Later) so the agent backs off rather than tight-looping. A node that already has a registered tunnel does not consume a new slot when it reconnects; the existing bridge is closed first. * fix(pilot): release backpressure as soon as the tunnel buffer drains Previously the bridge paused HTTP request bodies when the tunnel WebSocket's bufferedAmount climbed above 4 MB but never resumed them explicitly; the next data event re-checked the threshold, and TCP streams only saw a 'drain' fan-out on the 30 s ping cycle. Slow-consumer peers held buffered bytes for tens of seconds longer than needed. Replace both with an on-demand drain check: when at least one stream is paused, sample bufferedAmount every 100 ms; once it drops below the high-water mark, resume every paused request and emit 'drain' to every accepted TCP stream, then stop the timer. Dormant when no stream is paused, so steady-state cost is zero. * fix(pilot): trust internal CAs via SENCHO_PILOT_CA_FILE Self-hosted deployments often terminate TLS with an internal CA. The only previous escape hatch was NODE_TLS_REJECT_UNAUTHORIZED=0, which disables verification across every outbound connection in the agent process and is the wrong shape of fix. Add SENCHO_PILOT_CA_FILE: when set, the agent reads the file as a PEM bundle and passes it to ws as the `ca` option for the tunnel WebSocket. rejectUnauthorized remains true. Failure to read the file exits with a clear error so the operator does not silently fall back to the default trust store. There is no flag to disable TLS verification entirely; that would defeat the credential trust model. * fix(pilot): gate frame-rate-adjacent logs behind developer_mode The agent's malformed-frame warning fires per inbound frame and could flood logs under attack or against a buggy primary. The bridge swallowed parse errors silently, leaving operators blind to protocol drift, and the manager had no signal at all on tunnel registration. Route per-frame and per-tunnel diagnostic logs through the existing isDebugEnabled() helper (same pattern used by middleware/auth, RBAC, and websocket/logs). Production stays quiet by default; the toggle is the existing developer_mode setting in global_settings — no new env var, no new dependency. * feat(pilot): expose per-tunnel metrics via /api/system/pilot-tunnels Operator support had no signal beyond raw logs to answer "is this tunnel flapping?" or "is one bad node hiding behind the aggregate?". Add an in-process counter set covering tunnels_total, tunnels_replaced, tunnels_rejected_capacity, enroll_acks, frame_decode_errors, plus a per-node array carrying the connectedAt and bufferedAmount so a single tunnel sitting on a stuck buffer remains visible. Counters live in services/PilotMetrics.ts; the gateway has no shared in-process metrics facility today, so this is a per-feature pattern documented as such for future consolidation. Counters are strictly process-local — no telemetry, no export, no phone-home — consistent with Sencho's privacy posture. Surfaced read-only via GET /api/system/pilot-tunnels behind admin auth, mirroring the existing cache-stats endpoint. * fix(pilot): tighten reconnect backoff and log handling Three small follow-ups to the hardening pass: - Sanitize the primary URL on the agent's connect log so a malicious SENCHO_PRIMARY_URL with embedded control characters cannot inject fake log lines. - Move the reconnect-backoff reset from the 'open' callback to the 'hello' frame handler. A peer that always rejects the handshake (incompatible version, consumed enrollment token) used to reset the backoff on every TCP-level connect and tight-loop reconnects; now the reset waits for a clean protocol round-trip. - Document the StreamIdAllocator wrap behavior so future readers can see the relationship between the 2^31 wrap and the MAX_STREAMS_PER_TUNNEL cap without re-deriving it. * test(pilot): add coverage for enrollment, rate limiter, and tunnel caps Adds two test files exercising the hardening pass surfaces that were previously uncovered: - pilot-enrollment.test.ts: end-to-end through POST /api/nodes (pilot mode), POST /api/nodes/:id/pilot/enroll, the SHA256 token-hash persistence, replay protection, expired enrollments, and the rate-limit header wiring (enrollment limiter on pilot paths, global limiter elsewhere). - pilot-bridge-limits.test.ts: oversize-frame rejection at both binary and JSON decoders, and the per-tunnel concurrent stream cap as observed via openTcpStream returning null and the loopback HTTP server returning 503 once the bridge is at capacity. Bridge cap test fills the slot map with TCP stream handles rather than real HTTP requests so the OS socket pool stays out of the picture. * test(pilot): cover manager metrics snapshot and capacity-error shape Adds direct coverage for: - PilotTunnelCapacityError exposing the limit so the upgrade handler can format a useful close-frame reason. - PilotMetrics.snapshot returning a defensive copy (callers must not mutate the live counter set). - getMetricsSnapshot returning the open count, per-node breakdown, and counter set in a stable shape, with tunnels_total bumping on a successful registerTunnel. The manager test seeds a real pilot-mode node row first because updateNode throws on missing rows; this models the production path where a node is created before its enrollment is consumed. * docs(pilot): document tunnel limits, custom CA, and new failure modes Refresh the Pilot Agent feature doc to cover the hardening surfaces: - New 'Self-signed primary TLS certs' section walking through the SENCHO_PILOT_CA_FILE env var with an example docker run command. - New 'Resource limits' section listing the per-tunnel and system-wide ceilings so operators know what the wire enforces. - Four new troubleshooting entries: WebSocket close 1013 (system cap), loopback 503 (stream cap), close 1002 with protocol error (frame size / malformed), and HTTP 429 on enrollment. Also expand .env.example with a new pilot-agent block covering SENCHO_MODE, SENCHO_PRIMARY_URL, SENCHO_ENROLL_TOKEN, and SENCHO_PILOT_CA_FILE so operators do not have to read code or the feature doc to discover the agent-side config surface. * fix(pilot): address code-review findings on the hardening pass Critical: - Bridge: TCP backpressure now starts the drain timer and tracks streams awaiting drain, so a TcpStream caller waiting on 'drain' no longer hangs when no HTTP request is in flight. - Bridge: every direct streams.delete call now goes through the removeStream helper so per-stream idle timers are cleared consistently and pausedReqs / tcpAwaitingDrain stay aligned. - Bridge close(): resume any paused IncomingMessage before clearing the map so a parser does not stay stuck across teardown. High: - Protocol: decodeJsonFrame now compares Buffer.byteLength(raw, 'utf8') against the cap; raw.length (UTF-16 code units) let multi-byte payloads sneak ~3x the byte budget through. - Agent: the optional CA bundle is read once at construction and cached on the instance, so a missing or rotated SENCHO_PILOT_CA_FILE no longer process-exits on every reconnect attempt. Medium / Low: - Bridge limits test: assert mockWs.sent.length to confirm slot allocations actually serialized a tcp_open frame and that a rejected (cap+1) attempt did not consume a stream id. - Manager: defer the tunnels_replaced increment until after the cap check passes, so a rejected reconnect does not double-count as both replacement and capacity rejection. - Manager: clarify the protocol comment about MAX_FRAME_SIZE_BYTES enforcement layering (ws maxPayload is authoritative; decoder check is for tests and defense-in-depth). - Limiter: defensive `if (!req.body) return false` so a future refactor that delays body parsing cannot silently skip the enrollment limiter. Deferred (documented as follow-ups, not in this branch): - M2: narrowing public surface of PilotTunnelManager.getBridge to a MeshTunnelHandle interface (cross-cutting refactor). - H4: end-to-end replay test driving the upgrade handler twice with the same enrollment JWT (needs WS test harness).	2026-05-07 21:28:35 -04:00
Anso	8e7a567f69	feat: pilot agent outbound-mode for remote nodes (#667) ... * feat: pilot agent outbound-mode for remote nodes Adds a second mode for managing remote nodes: the agent dials an outbound WebSocket tunnel to the primary, so the remote host no longer needs an inbound port, a reachable URL, or its own TLS certificate. Works behind NAT, residential routers, and corporate firewalls. The primary multiplexes HTTP and WebSocket requests over a single tunnel via a hybrid JSON + binary frame protocol, bridged through a per-tunnel loopback server so existing proxy and upgrade handlers route pilot-mode nodes identically to proxy-mode ones. Enrollment uses a single-use 15-minute pilot_enroll JWT exchanged for a long-lived pilot_tunnel credential on first connect. Proxy mode continues to work unchanged and both modes are supported side-by-side. * test(e2e): switch to proxy mode before asserting api_url field Remote nodes default to Pilot Agent mode, which hides the api_url input. The SSRF-validation tests need proxy mode, so the helper now selects Distributed API Proxy after picking Remote type before asserting the field is visible. * fix(e2e): wire Combobox id prop so node-mode selector resolves The Combobox trigger button had no id, leaving its Label orphaned and making getByRole name-based lookups fail. Adding id to the primitive, passing id="node-mode" from NodeManager, and updating the E2E helper to use #node-mode fixes both the a11y regression and the CI timeout.	2026-04-17 20:31:43 -04:00