Matthew Meszaros
de8da6396e
feat: build the tester sign-in address from DASHBOARD_URL instead of rewriting the admin origin's first label to dev., which handed an OAuth reviewer a different deployment than the dashboard every other link in the panel points at
2026-09-16 19:40:14 +02:00
Matthew Meszaros
03f813dd43
feat: let an admin create a tester account that joins an existing workspace with an explicitly chosen role instead of always minting an empty one, so an OAuth reviewer signing in lands in a workspace that shows the app doing real work
2026-09-16 19:37:32 +02:00
Matthew Meszaros
0334d2c64a
feat: prevent ResizeObserver feedback warnings from reaching PostHog and scope worker outage alerts per workspace on shared cloud workers
2026-09-16 17:42:38 +02:00
Matthew Meszaros
140c7de436
feat: add the admin panel's Promo codes page and route the six /admin/discounts endpoints that existed as handlers but were never wired, so a launch offer is built in the operator UI instead of an INSERT against production, with caps that an explicit null can actually clear on PATCH
2026-09-16 04:04:09 -07:00
Matthew Meszaros
ae012dd13f
Clear the live error-tracking issues, and the workspace rename that renamed the wrong workspace ( #533 )
...
* feat: stop a managed Kafka cluster refusing topic creation from failing the publish, by treating a topic- or cluster-authorization failure from CreateTopics as a topic the cluster owns rather than one that is missing, which on Confluent Cloud dropped every warmup event and filed one issue per message because the topic never became known
* feat: drop a report whose error is a cancelled context in errs rather than at ninety call sites, so a browser navigating away or a container draining on deploy stops filing one issue per query that happened to be in flight, while a deadline this process set and blew through still reports
* feat: stop renaming one workspace from renaming another, by keying the workspace settings editor on the workspace id so a switch re-seeds the name field instead of leaving the previous workspace's name against the new workspace's autosave baseline, and pinning every save on the workspace, sending and tracking pages to the workspace its draft was hydrated from
* feat: drop Script error. and the ResizeObserver notice on the marketing site and the hosted form page the way the dashboard and admin panel already do, since those two carry no stack and no bug and between them were the largest issues in error tracking, all of it from warmbly.com
* feat: rename the forms Turnstile script module to turnstileScript.ts so it no longer differs only in case from the Turnstile.tsx component, which resolved both imports to one file on a case-insensitive filesystem and failed forms' typecheck with TS1149
* feat: upload source maps from the static build:pages build as well as the image build, so the dashboards served from a static host stop reporting every stack frame as a minified name beside 'Invalid source map: bad json', which is PostHog falling back to fetching the .map from a host that answers with its SPA fallback
* feat: build every admin list in pg_admin.go with make rather than declaring it nil, so an empty page serializes as [] instead of null, and guard the audit table's own empty check, which is what crashed admin.warmbly.com/audit with 'null is not an object (evaluating d.data.length)' whenever a filter matched nothing
* feat: match the whole broker description rather than a substring when deciding a topic create was refused for permissions, since that answer remembers the topic as present, and clear the cached promise and dead tag when the forms Turnstile script fails to load so a blocked first attempt no longer leaves every later mount with the same rejection and the captcha permanently missing
2026-09-15 09:05:53 -07:00
Matthew Meszaros
179af5815f
feat: type the four AdminUserPreview slices as nullable in the admin panel's model so a backend older than the empty-slice fix in pg_admin.go cannot crash a page through an unguarded email_accounts.length again
2026-09-15 13:39:05 +02:00
Matthew Meszaros
0a1ed6f04e
feat: resolve scanner ASNs from a GeoLite2-ASN database so the catalogue's asn: entries match without a Cloudflare transform rule, ship the Proofpoint, Mimecast and Cisco ASNs enabled behind a new probable certainty that widens the consumer's machine window instead of deciding the verdict, make the tracking event dedupe claim one coalesced operation, and report an ASN database that opened cleanly but resolves nothing ( #440 )
2026-09-15 01:48:45 -07:00
Matthew Meszaros
92e298b6ec
fix: clear every open issue in error tracking by fixing the bugs behind them rather than the reports: a document-level mouseleave handing RippleProvider the document itself, whose classList is undefined; the admin panel posting /getaway without the /v1 its baseURL omits, so its realtime socket 404d on every page; Gmail throttles classified from the 403 status alone and told to re-authorize instead of back off; consumer flag and folder events retrying forever on a message the unibox never stored; a lost token-refresh race answered 500 instead of the documented 401; a nil email_accounts slice crashing the admin user page; Turnstile mounted with an empty sitekey; conditional passkey autofill run after the user navigated away; a boot log filed as an issue; and one publish failure per message on a topic the broker refuses ( #519 )
2026-09-14 21:06:14 -07:00
Matthew Meszaros
d74d5e6836
fix: retire the warmup spam score, a counter that grew with volume rather than misbehaviour and that no band could act on ( #508 )
...
* fix: retire the warmup spam score, a ratchet that grew with volume rather than misbehaviour and that no band ever read, dropping the column from the pool row and the reputation ledger and explaining a pool finding with the band's own reason instead (#491 )
* test: pin the advisor snapshot's pool columns against the scan, since the band's reason now reaches the finding through that select alone (#491 )
* fix: hold a warmup sentence's score and reason with the sentence itself, keep the retired spam_score key on the published analytics payload as a deprecated zero, seed the sandbox with severity-shaped scores, and record the raw spam report when the warmup service is absent (#491 )
2026-09-14 07:44:34 -07:00
Matthew Meszaros
7300b3b021
feat: full PostHog coverage: identify the signed-in user and workspace in the dashboard and admin panel with autocapture, heatmaps, dead and rage clicks, web vitals, network timing, console capture and session replay masking only password fields, send server-side signup, trial and subscription events under the user id with the organization as a group, keep the marketing site and form pages cookieless while capturing everything stateless plus a form funnel, upload the form app's source maps, and add WARMBLY_POSTHOG_SESSION_REPLAY
2026-09-13 20:58:22 -07:00
Matthew Meszaros
1dc4aedc3c
fix: retire the warmup invalid-token band with its table, metric query, service and repository methods and admin tab, since nothing has fed it since #481 and no attributable forged-token signal exists; key the live pool fixtures on the canonical pool ids so the warmup, repository routing and tasks routing suites run on a fresh database, and correct every doc, site and advisor line that still described the retired signal or a spam-score threshold nothing implements ( #490 )
2026-09-13 08:09:01 -07:00
Matthew Meszaros
8799680166
fix: never charge a mailbox for a warmup token that arrived in its inbox, hold a quarantine or block for its full term against fresh metrics, and keep a penalised address's standing across removal, pool exit and export through a trigger-maintained mirror, since the recipient never controlled the token, the bands read seven days against 30-day terms, and the pool row died on paths a snapshot at deletion never saw ( #481 )
2026-09-13 04:34:44 -07:00
Matthew Meszaros
43dcbde06c
feat: tester accounts, creatable from the admin panel ( #483 )
...
* feat: excuse one named account from the emailed login code, so a vendor reviewer who cannot read this instance's mail can sign in without turning codes off for everyone, with the reason recorded beside it and every run of warmblyctl status naming the accounts that hold one
* feat: create and manage tester accounts from the admin panel, so letting a reviewer in is a form rather than a shell, with the password shown once and every live exemption listed on one page because forgetting one is the way this goes wrong
* fix: give tester management its own permission bit rather than borrowing ban_users, create the account and its exemption in one transaction so no invisible orphan survives a failure, require an accountable operator on the CLI grant, stop a halted row scan reading as the whole exempt list, and show a failed query as an error instead of as no testers
* chore: re-run CI after the aggregator tripped on a cancelled job from the branch update, with every underlying job green
* chore: retrigger CI, the previous run sat queued indefinitely while other branches ran
* feat: roll back a half-created tester when its workspace step fails and backfill the manage-testers bit onto admins already holding every other permission, so the address is not left taken by an unusable account and the new routes are not 403 for the existing admin
* feat: make the 000151 manage-testers backfill one-way, because clearing bit 22 on the way down would also revoke it from an admin granted it explicitly afterwards and the up migration would not restore that
2026-09-13 03:07:10 -07:00
Matthew Meszaros
7f74664c72
fix: a granted plan unlocks nothing, and Turnstile never renders ( #474 )
...
* fix: let a granted plan unlock the dashboard, since the client decided paid from the Stripe status a managed subscription never touches, and replace the Turnstile size Cloudflare removed so the widget renders and can issue a token at all
* feat: tell people on a preview deployment that it is a public beta, once in a dialog and thereafter as a header pill they can reopen, driven by a config value rather than a hostname so one image stays reusable, and bind both Turnstile modals through onLoad because the component is not forwardRef and execution=execute never fires without the widget instance
* fix: keep the beta pill outside the desktop-only header group so the notice stays reopenable on a phone, and say in the docs that the value is baked into config.js at container start rather than read per load
2026-09-12 21:45:05 -07:00
Matthew Meszaros
017f4cf60f
feat: plans an operator can grant, visible in the admin panel ( #467 )
...
* feat: add operator-granted plans so a workspace can be paid without Stripe, surfaced in the admin panel as a badge, a filter and a card carrying who granted it and why, because the only alternative was writing a fake stripe subscription id into the database
* fix: hold a granted plan beside the paid one rather than over it so a Stripe workspace returns to the plan it pays for when the grant ends, route entitlement lookups through EffectivePlanID, separate a repository failure from an unknown plan, end a grant at local end of day, and drop an index that served no query
2026-09-12 09:45:31 -07:00
Matthew Meszaros
47defafa09
feat: fix the six self-host defects reported in issue #439 ( #456 )
...
* feat: fix the six defects reported in issue #439 by mapping the IMAP UNAVAILABLE, INUSE and NONEXISTENT response codes to retry-level errors instead of a critical reconnect prompt, synthesising a stable no-msgid key so one message with no Message-ID header can no longer 400 the internal map endpoint and wedge every later sync pass with its cursors held, adding mailhtml.FromText and HasContent so an API or agent-created step with a plain body stops shipping the composer's empty div placeholder as its text/html part (derived on create and plain-only update, exposed as body_html on update_campaign_step, dropped at send and preview time, and refused at campaign start with empty_step_body), honouring sender_strategy='explicit' in ResolveCampaignSenderPool and ValidateCampaignReady so an emptied explicit pool parks the campaign instead of widening it to every mailbox in the workspace, making the paused_no_accounts auto-pause loud with an error log line, an error-level activity-feed entry and an org-scoped CAMPAIGN_PAUSED realtime pulse, gating the admin sign-in's Turnstile widget on GET /v1/auth/config so a self-host with CAPTCHA_PROVIDER=none is not locked out, and parsing NATS_URL down to its host:port so a credentialed bus URL no longer reports NATS down
* feat: act on the self-review of the issue #439 fixes by dropping the campaign wizard's own escapeHtml body_html builder, which entity-escaped the quotes in a conditional and made the template fail to parse at send time, and letting the backend's FromText render that part instead so wizard-written steps also get their bare URLs linked for click tracking, correcting the docs and openapi description that claimed an explicit sender pool never falls back when it still unions its tags as migration 000013 designed, extracting the duplicated blank-HTML-part guard into dropBlankHTMLPart shared by the send path and the preview, and recording why the no-msgid key keeps the folder name despite a RENAME changing it
* feat: address the CodeRabbit review on the issue #439 fixes by holding the admin sign-in's Turnstile widget unmounted until /v1/auth/config resolves so an instance with no route to Cloudflare cannot raise a widget error on a screen nobody submitted, failing StartCampaign closed when the sequence read errors rather than skipping both the malformed-template and empty-body refusals, giving TCPCheck the default port its protocol assumes so a portless NATS_URL is no longer reported down, leaving a URL that carries a merge field unanchored because the send path renders bodies with text/template and a quoted contact value would break out of the href, and correcting the sequences guide and the Campaign and CampaignUpdate openapi descriptions that named the wrong tag field
2026-09-12 03:13:38 -07:00
Matthew Meszaros
80c3c79a31
feat: add live coverage that the machine windows survive the settings document's jsonb round trip and reach the classifier, a guard that every shipped scanner CIDR is written as its own network address since the loader truncates host bits silently, and correct the comments that claimed an edit lands on the very next event when the consumer reads through a thirty second cache in its own process
2026-09-11 03:11:35 -07:00
Matthew Meszaros
184a3dc08e
feat: make the automated-open and automated-click windows operator-editable under Instance settings and raise their defaults to 60s and 30s, because the ten-second window was anchored on dispatch to the worker rather than on delivery and routinely expired before the recipient-side gateway it was meant to catch had even seen the message, and add Barracuda's published Email Gateway Defense blocks to the scanner catalogue with Proofpoint, Mimecast and Cisco shipped commented out because browser isolation renders a clicked page from the vendor's own network
2026-09-11 02:53:34 -07:00
Matthew Meszaros
170c33780a
feat: pass ui_host to PostHog in the dashboard, admin panel and marketing site from its own environment variable so a proxied api_host stops breaking toolbar and session-replay links, which the SDK builds against whatever it sends events to and which a reverse proxy does not serve, defaulting to us.posthog.com so an install that does not proxy is unaffected
2026-09-11 11:34:02 +02:00
Matthew Meszaros
ced741e352
feat: make PostHog the default error tracker across every runtime while keeping Sentry fully supported alongside or instead of it, by turning internal/observability/errs into a two-sink fan-out with a local-log fallback, adding $exception capture to the Go services, the Rust tracking service, the Elixir realtime service and the dashboard, admin and form apps, reporting gin panics with their route, request id, workspace and user, attaching that identity plus a route and failed-request trail to browser exceptions, and wiring POSTHOG_ERROR_TRACKING, the node join env, compose, source-map upload and the docs to match
2026-09-10 19:11:32 +02:00
Matthew Meszaros
6ebf9cfdcf
Merge remote-tracking branch 'origin/main' into fix/self-hosted-unsubscribe-domain
2026-09-09 08:24:04 -07:00
Matthew Meszaros
7d58b874b8
feat: keep every recipient-facing and self-host-facing address on the deployment's own domain: mint unsubscribe links on a workspace's verified tracking domain (served by the tracking service, proxied to the backend that owns the pages), attach RFC 8058 one-click only over https, resolve all branding through config.Brand() gated on SelfHosted() so a self-host's email footer, sign-in links, stats card, API example and public form badge name nobody else, drop the app.warmbly.com fallback from AppBaseURL, blank TRACKING_DOMAIN and FORMS_DOMAIN on core-only installs, and have install.sh offer to configure a fresh interactive install instead of silently defaulting to localhost
2026-09-09 06:34:43 -07:00
Matthew Meszaros
0a08983158
feat: keep the node's write path reachable and the image reference root-owned, by granting traverse on the state directory without making it writable, re-owning image-ref on every join so a machine joined by the earlier chown-everything script converges, mounting BLOB_FS_ROOT separately when blobs live under the state directory the mount no longer covers, and renaming the reserve dialog's remaining convert-to-dedicated button
2026-09-09 05:53:28 -07:00
Matthew Meszaros
65fe82c472
feat: stop the node container being able to choose the image root runs, by giving it a dedicated uid-1000 subdirectory to write its update target into instead of making the whole state directory writable, validating the tag it asks for before the updater acts on it, and no longer recursively chowning a directory a bare-metal install also keeps its blobs in
2026-09-09 05:41:16 -07:00
Matthew Meszaros
bfdbd77b04
feat: fix the eight defects the second review pass found, including three where the previous fix did not land: the bind-mounted state dir was root-owned so the node running as uid 1000 still could not write its update target, elevating the reserved-worker eviction did nothing because the rotation loop bailed on target-equals-current before the urgency was consulted, and the warmup-pool assertion was vacuous which hid that warmupPoolFor checked the subscription repo before the billing provider and answered free on a self-host install
2026-09-09 05:32:30 -07:00
Matthew Meszaros
8b83062eff
feat: fix ten defects the review found, the worst being that every seeder still wrote the worker columns migration 000141 moved to fleet_nodes so make dev could not seed, that the generated systemd unit used a command substitution systemd never expands so a joined machine restart-looped while the script reported success, that the state directory was not bind-mounted so the node wrote its update target inside the container where the host timer never saw it, and that placement stopped assigning warmup pool membership so paid mailboxes silently warmed in the free pool
2026-09-09 05:21:17 -07:00
Matthew Meszaros
435dbb522f
feat: replace the worker tier/type/risk-pool/egress categories with a scored placement model and make the fleet pull-based, so a machine joins with one command, workers and consumers share one node registry with usage and liveness, nodes self-update to the version the control plane resolves, and the Hetzner provisioning, worker profiles and SSH orchestrator are removed
2026-09-09 04:54:01 -07:00
Matthew Meszaros
c8ecf84e5e
feat: report the highlighted search-picker option only through aria-activedescendant and keep aria-selected false until an option is actually picked, as the CodeRabbit re-review on the admin frontend PR asked
2026-09-07 23:16:18 -07:00
Matthew Meszaros
f9bef0d5f1
feat: address the CodeRabbit review on the admin frontend PR by warning about the last admin who can grant access only when the whole admin list is on screen, resetting the convert-to-dedicated form on Cancel, following a changed ?q= on the mailboxes page instead of only its first value, treating the all sentinel as the sync state default and not as a search term, naming the lease window in the webhook reclaim confirmation only once it is loaded, walking transfer group dependents transitively so unticking a group drops everything that needs it, labelling the shown-page scope of the worker bulk move, closing the mobile drawer when the viewport reaches the desktop breakpoint, dropping the stale plan mention from the users page description, and completing the accessibility contract of the status segments, passphrase fields, icon-only delete button and search combobox
2026-09-07 23:08:56 -07:00
Matthew Meszaros
800c4a062f
feat: frontend half of the admin panel upgrade: drop ten unused dependencies, the dead worker-load and plan requests, the stub pages and the retired permission names, fix the analytics client that sent the wrong query parameters and the mail test toasts that used an unmounted toaster, add a Cmd-K command palette with live user, organization, mailbox and worker search, a mobile nav drawer, document titles, a Sentry-wired route error boundary and permission gates on every route, replace polling with the realtime spine wherever an event exists, fold Analytics into Overview with a signups-by-channel card, merge System Status into Setup and health and Settings, Notifications and Effective limits into Configuration as tabs with redirects from every old path, add the Sync, Sends, Jobs, Fleet, Admins and Transfers pages plus API-key, webhook and transfer tabs on the organization page, mailbox reassignment on the worker page and abuse and action-history tabs on Warmup, and add the admin-panel docs page with every nav reference updated
2026-09-07 21:40:38 -07:00
Matthew Meszaros
26c88d2426
feat: address the CodeRabbit review on the PostHog PR by dropping a referrer that is not a hostname instead of storing its query string or fragment, redacting email-shaped values out of every acquisition field before they reach the database or an analytics property, counting an invited signup which returned before the count was taken, hanging subscription_started off the persisted trial-to-paid transition so a redelivered webhook cannot report a second start, capturing the validated provider in both OAuth mailbox paths including Warmbly Cloud, making the two acquisition toggles mutually exclusive and naming them after what they actually select, and replacing the unsupported CNIL consent-exemption claim with what the guidance says and an explicit note that qualifying is a deployment-specific assessment this document does not make
2026-09-07 05:45:11 -07:00
Matthew Meszaros
8e9f67f46b
feat: add cookieless PostHog analytics for the hosted marketing site and dashboard with server-side signup and subscription events, a first-party acquisition record written once at signup on a new organization_acquisition table registered in the org-transfer spec, an acquisition column and channel filter in the admin org list, and never a single request from a self-host because every key is unset by default
2026-09-07 04:18:33 -07:00
Matthew Meszaros
51dedc90ee
feat: put every runtime behind one optional error-reporting story: a single internal/observability/errs wrapper that is now the only package importing sentry-go, InitSentry for cmd/forms, release and environment tags on every service from the existing build stamp, optional Sentry in the admin panel and the public forms app, the sentry crate in the Rust tracking service, release tagging in realtime, CI source-map upload that only runs when a Sentry token is configured, and docs covering the DSN for each service
2026-09-07 03:51:06 -07:00
Matthew Meszaros
b0c090e521
feat: keep an admin-cancelled warmup generation batch job running with batch_status cancelling instead of marking it failed on the spot, so PollBatches reaches the cancelled branch that PR 333 added and ingests the conversations OpenAI finished before the cancel landed while the admin's reason survives the ingest, refuse a second cancel on a batch already cancelling, hide the admin Cancel button while a batch is cancelling, and refresh the PollBatches, CancelBatch and WarmupBatchPoller doc comments that still said expired and cancelled batches are marked failed
2026-09-05 21:14:46 -07:00
Matthew Meszaros
5fd9aedb80
Merge remote-tracking branch 'origin/main' into feature/plan-upgrade-paywall-modal
2026-09-04 09:11:51 -07:00
Matthew Meszaros
b0db8cc02b
feat: chain the backup bundle cleanup behind a successful docker compose cp so a failed copy cannot delete the only backup, and tell a pinned image install to set WARMBLY_TAG before pulling by hand
2026-09-04 06:35:04 -07:00
Matthew Meszaros
d78e4423e4
fix: address the review of the notifications work — bound detached alert delivery to a fixed pool that drops and logs rather than spawning a goroutine per event, never echo a transport error that embeds the webhook URL, treat an emptied channel target as cleared so switching transport cannot post Slack payloads to a Discord URL, emit user.registered for invited signups and the worker alert after a successful reassignment, format Stripe amounts by currency exponent, refuse an upgrade while the plan list is still loading instead of silently opening the portal, apply a fixed discount to the annual invoice rather than to each month, convert proration from minor units, contain focus in the nested enterprise dialog, hide the odometer wheels from assistive tech, and drop the last dedicated-IP row from the pricing matrix
2026-09-04 06:33:20 -07:00
Matthew Meszaros
77058a2cb4
feat: address the review on the installer branch by keeping the database password out of pg_dump's argv, excluding backup bundles from the blob root they are written into, tolerating blobs that change or vanish mid-archive, making the instance-settings bootstrap a single atomic insert, and validating the release tag before it is written into .env
2026-09-04 06:24:05 -07:00
Matthew Meszaros
a0c9d5a5b0
feat: add operator notification channels an admin configures in the panel (Discord, Slack, HMAC-signed webhook, or email) with per-channel event subscriptions, a synchronous test probe, SSRF-guarded URLs and credentials redacted on read, wire nine instance events including the enterprise inquiry that previously wrote a row nobody was told about, and reword the plan-specific limits copy so it reads correctly without billing
2026-09-04 06:02:40 -07:00
Matthew Meszaros
d68bbcd2ab
feat: add a one-command self-host installer at warmbly.com/install.sh with an interactive data-control wizard, give docker-compose.yml image keys and per-store volume variables, add an image-mode updater, move engagement/form/audit retention into instance settings, and add warmblyctl backup/restore
2026-09-04 05:49:54 -07:00
Matthew Meszaros
3d5eb36ae5
feat: second review pass on self-hosted updates: git describe suffixes after a prerelease (rc.1-2-gabc1234) now parse as prerelease plus commit distance, Apply validates the updater for every target including explicit tags, a DNS failure only reads as the compose profile being off when UPDATER_URL names the updater service and stays unreachable for custom hosts, the admin dialog re-checks that the update can still start before launching it and drops the confirmation when it cannot, the bare-metal installer refuses symlinks anywhere inside a build directory, restores config.js with --remove-destination and bounds every health probe, and the bare-metal docs add the single sudoers rule before the upgrade command
2026-09-03 05:04:30 -07:00
Matthew Meszaros
bb35de0455
feat: address review on self-hosted updates: the updater no longer re-locks its mutex when a job finishes (every job used to deadlock at completion and freeze the status API), the backend caches the updater view so the member version pill, the health checks and the admin poll share one read and an absent updater is reported as not running rather than broken, the bare-metal upgrade builds unprivileged and hands off to a root-owned fixed-path installer that refuses symlinks so sudoers allows one command instead of install/cp/rm/chown/chmod/systemctl/ln, the installer fails when the backend does not come back, the seed image gets the version build args, the dashboard gates the update action on manage_settings and stops polling a backend that answers 404, and revived timestamps are typed as Date
2026-09-03 05:04:30 -07:00
Matthew Meszaros
b2ea1f1961
feat: add self-hosted update awareness and one-click updates: every binary is stamped with its version and commit, the backend polls GitHub Releases and a new host-side updater (cmd/updater, compose profile or systemd unit) reports the checkout's commit distance, the admin panel's top bar shows a version pill that turns into an update indicator and opens a dialog with confirmation, live step progress and log, restart tracking and result, the dashboard header shows the same pill to every member of a self-hosted instance with the full update flow for platform admins, Setup and health gains update_available and updater_unreachable checks, warmblyctl status prints the version, make upgrade and scripts/upgrade-bare-metal.sh cover the by-hand paths, and docs gain an Updates page plus configuration, health, deployment and API reference updates
2026-09-03 05:04:30 -07:00
Matthew Meszaros
a02ff7c936
feat: address verification overhaul for #264 : MillionVerifier as a pay-as-you-go integration plugin with org-sealed key and automatic fallback to the built-in check, built-in prober gains domain cache, Microsoft/Yahoo fingerprinting, MX fallback, disposable/role sub-statuses and a self-check breaker, imports and POST /contacts accept verification results in any known provider vocabulary with auto-detected columns, verdict provenance and expiry columns (migration 000110), campaigns park at paused_undeliverable with re-verify/send-anyway instead of finishing, POST/GET /contacts/verification bulk actions and overview, launch gate override via acknowledge_list_risk, animated verification marks, banner and settings card in the dashboard, and docs
2026-08-29 23:11:22 -07:00
Matthew Meszaros
2fd22cafd3
feat: answer 404 rather than 500 when a risk mutation names an organization that does not exist, by having every org-risk repository path return nil for a missing row and one resolved() helper turn that into the not-found the GET already gave, and stop showing an admin who holds only view_organizations the Set posture, Lift override and retract controls, which the backend can only ever answer 403, so the Abuse posture panel reads as evidence for them and as evidence plus actions for an admin who can actually act
2026-08-28 22:47:24 -07:00
Matthew Meszaros
a5ea55bba5
feat: give a suspended workspace a way back, because risk_state was a one-way door: the derived band is no longer pinned at suspended by the UPDATE in pg_org_risk, an operator's decision is now an explicit risk_override that outranks the score and survives every later detector write until it is lifted, the one-shot detectors (signup origin, import list quality, repeated sign-in anomalies) file findings with a 30-day expiry that a six-hourly consumer sweep retires so a score falls on its own, migration 000104 backfills that expiry onto findings already on file, and four admin endpoints plus an Abuse posture panel in admin/ let an operator finally read the evidence the customer endpoint withholds, retract a finding, pin a band and lift the pin; also stops the reviewing admin's identity reaching the tenant's own audit feed (which resolves an actor to a name and email) by recording the platform as the actor there and the operator in the admin trail, and stops risk_signals riding along in a customer-downloadable org export
2026-08-28 22:42:57 -07:00
Matthew Meszaros
0f715da88e
feat: close the force-stop race and trim the comments the review flagged: the eligibility test moved out of the admin service and into the UPDATE's WHERE clause, so a campaign that completes between the status read and the write can no longer be dragged back to paused and recorded as force-stopped, StopCampaign reports whether it actually stopped anything and the service turns a refusal into the same 400 it used to raise from the pre-check, a new live test parks the campaign at completed and at draft and fails if either is overwritten, and the explanatory comment blocks added across pg_admin.go, the admin service, the admin models and the rate-limit dialog are cut back to the one-line form CLAUDE.md asks for
2026-08-27 03:21:25 -07:00
Matthew Meszaros
1c4fcff558
feat: make the admin panel's broken queries run: seven statements referenced schema that does not exist and failed 100% of the time, so a force-stop wrote status = 'stopped' and stopped_at to a campaign_status enum and a campaigns table that have neither, the plan writes named a duration column that became duration_id long ago, the user rate-limit read and write named a daily_email_limit column that user_rate_limits has never had, and the user preview compared email_accounts.user_id (uuid) against a text parameter and then swallowed the error so every operator saw an empty mailbox list; the same uuid = text defect in GetUserEmails, which the prepare sweep cannot see because that WHERE clause is assembled at runtime, was a live 500 on GET /admin/users/:id/emails; a stop now parks the campaign at 'paused' like the owner-facing stop and records the reason the UI has always sent and the backend has always dropped into both the audit log and the owner's campaign feed, the plan writes resolve durations.title to duration_id and answer 400 rather than a constraint violation on an unknown period, the rate-limit editor now covers the seven real limit columns instead of one that never existed and patches insert-then-update in a transaction because every column is NOT NULL, AdminWorkerEmail.LastSyncedAt is a pointer so a never-synced mailbox stops being silently dropped from every admin list, and TestLiveEveryQueryPrepares now fails on undefined columns, tables, operators and enum values instead of only reporting them
2026-08-27 03:09:40 -07:00
Matthew Meszaros
42ddb78e60
feat: make the realtime service survive a client that spams channel joins, and stop both clients from being that client: ws_join is now spent per phx_join in a shared RealtimeWeb.ChannelGuard that every channel runs BEFORE its Auth lookup (it was only ever spent on the socket handshake, so an established socket could issue unlimited joins and each one cost a Postgres query), handshakes move to their own ws_connect bucket so a reconnect storm cannot eat the budget a client needs to rejoin with, the retry_after_ms hint now points at the next window instead of shrinking as overage grew, every channel's join rejection carries the numeric code the API reference already promised (plus 4005 for a malformed topic), BulkChannel only pushes an event to the user who owns the operation instead of to anyone who guesses the id, and the web and admin clients treat a rate_limited join as transient (wait out the hint, keep the socket) while backing off per-channel rejoins on a decaying schedule rather than reconnecting on the 120ms floor forever; adds an ExUnit suite for realtime and a vitest suite for admin, both wired into CI
2026-08-26 21:14:34 -07:00
Matthew Meszaros
3739a36b67
feat: enforce the persisted SPF/DKIM/DMARC state as a real cold-send and warmup gate behind a 72h grace clock and an operator toggle, after first fixing the DMARC organizational-domain fallback in dnsauth so a dedicated sending subdomain covered by its parent's record stops reading as unauthenticated, adding auth_state to the four mailbox loaders that never selected it (which would have made the gate dead code), stamping auth_failing_since on entry to failing so a resolver hiccup can never stop a campaign, notifying the org on that transition, and reporting an all-gated pool as ErrDomainAuthFailing instead of a message about sending windows ( #160 )
2026-08-22 09:37:26 -07:00
Matthew Meszaros