Xplorr Changelog

Changelog

Recent updates to Xplorr. For the full changelog with older entries, see the website changelog.

October 8, 2026: billing is live

Plans

New organizations start on the Free plan: free forever, 1 cloud account, 2 users, no credit card, no AI features.
Growth is USD 49 per organization per month, billed monthly in advance by card through Stripe, and includes the AI features. Cancel any time; Growth stays on to the end of the paid period, then the organization moves to Free.
Enterprise is custom and invoiced.
Taxes are not included; they are added at checkout where required.
Organizations that joined during the private beta keep their current plan at no cost, with at least 30 days notice by email before that changes.

Billing in the console

Admins upgrade from Billing in the sidebar (Admin section), or from the Upgrade button in the top bar on the Free plan. Manage billing opens the Stripe portal for the card, invoices and cancellation. See Billing and Plans and Limits.

October 7, 2026: inventory state, network periods, quieter email

Runtime state for more Azure resources

SQL databases, PostgreSQL and MySQL flexible servers, Container Apps, Application Gateways and VM scale sets now show Running or Stopped in Inventory. They showed N/A before.
Azure runtime state is read every 3 hours instead of once a day, and every status shows when it was captured. See Resource Inventory.
Types with no runtime state read N/A (no runtime state).

Network Costs by calendar month

Network Costs takes a custom start and end date, so a month can be checked against the provider’s bill. The presets cover complete days only, like the other cost pages. See Network Costs.

Sync errors you can act on

Click the sync error badge to see each failing account, the cause, when it failed and the last successful sync, with a Retry sync button. The badge counts only active accounts you can see. See Data Health.

Sync result emails are off by default

The email after every sync run is now opt in. Turn it on under Profile, Notifications if you fix sync problems. See Notifications.

September 28, 2026: any date range, fewer pages, complete days only

Custom periods in Cost Analysis

Compare any start and end date with the previous period of equal length, the same days last month, or a second range you pick. The summary cards, spend by provider, daily trend, spend by service, service table and account breakdown all follow the period, which is kept in the page address and in saved views. See Cost Analysis.
Savings Coverage takes a custom start and end next to its presets.

Fewer pages, nothing removed

Compare is the Accounts tab of Cost Analysis, since its period comparison is now part of Cost Analysis.
Teams and Tags is the Rules tab of Cost Allocation.
Savings Coverage is the Coverage tab of Commitments.
Approvals and Realized Savings are tabs of Recommendations.
Every old address opens the new tab.

Totals count complete days

Providers are still writing the last two days of cost, so totals now stop two days back: month to date, forecasts, comparisons and budgets alike. Preset windows such as “last 7 days” cover exactly that many complete days, and the cards show the dates.
Daily and weekly scheduled reports end two days before they are sent, and the monthly report is sent on the 3rd so the month’s last day is complete.
Azure Defender for Containers image scans and scale set instances show readable names instead of an image digest or a bare number.

September 28, 2026: virtual tags, bill checks, security keys, commitment history, AI budget

Virtual tags

A cost category exposed as a virtual tag now works as a dimension wherever cost is grouped or filtered. Rules are applied when you view the data, so nothing is re-synced when a rule changes.
Rules can test the resource ID and match a regular expression (ignores case, up to 200 characters, no backreferences, lookaround or nested repetition). Rules run in priority order and the first match wins; spend no rule matches is Unallocated, a name no value can take.
The rule editor previews each rule over the last 30 days, in its place among the other rules, before you save it.
A Cost by virtual tag card on Cost Analysis, budgets and unit metrics scoped to one virtual tag, and CSV and Excel exports with the virtual tag as a column. See Cost Allocation.

Bill reconciliation

Every night each AWS, Azure and GCP account’s total is compared with the provider’s own bill for the month to date and the previous month. Data Health shows Matches your AWS bill with both totals, Does not match with the gap, or Could not check with the reason, and flags cost that may be counted in two connected accounts. See Data Health.
An AWS Organizations management account no longer counts members that are connected separately a second time. Its cost now leaves them out, and Cloud Accounts says how many linked accounts are included and excluded. See Connect AWS.

Security keys

Register up to 10 security keys, choosing Security key (a YubiKey or other USB or NFC key), This device, or Passkey from a password manager or phone before the browser prompt. Mandatory 2FA setup offers a backup key right after the first. Keys can be renamed, any registered key works at sign in, and your only second factor cannot be removed. See Security keys and passkeys.

Commitments

Daily utilization and coverage trend per family and service, daily history under each commitment, and a 12 month expiry calendar highlighting commitments ending within 30 and 90 days.
Azure reservation and savings plan utilization now comes from amortized cost with Cost Management Reader, so it no longer depends on the tenant scope reader roles. See Commitments.

AI budget and plans

Every organization has a monthly AI budget, USD 10 by default. Admins are warned at 80 percent by email (the optional AI budget warning) and Slack; at 100 percent AI features pause until the 1st and everything else keeps working. When paid plans launch, the Free plan has no AI features. See Plans and Limits.

Signup

A new organization opens the console once it connects a cloud account or Kubernetes cluster, and has 14 days to do so, with reminders on day 7 and day 12. After that it is suspended, never deleted. See Quick Start.

September 23, 2026 — Credit tracking, audit log filters, mandatory 2FA

Credit balances

Track credit grants against a cloud account: amount, currency, start and expiry date, source (AWS promotional, AWS Activate, Azure sponsorship, Azure MACC or prepayment, Google Cloud promotional, or other) and a note. A new Credits page under Cost Management lists every grant with its remaining balance, a 30 day burn rate, a projected run out date, and the balance left unused if nothing changes before expiry.
Azure EA and MCA grants can read their balance directly from Azure instead of only from synced credit lines, refreshed on save, on demand and once a day. If the connected credentials cannot reach the billing account, the grant says which role is missing.
Alerts fire 30 and 7 days before a grant is projected to run out, and 30 and 7 days before expiry when balance will be left unused, by email and as a Slack card in the same branded layout as other alerts. Editing a grant’s amount, dates, source or currency re-arms its alerts.
Overlapping grants on one account draw down earliest expiry first. A credit line in a currency that does not match the grant is left out of the balance rather than summed, and the grant says so.
A viewer whose account access is restricted never sees a figure that depends on an account they cannot see; the grant reports that its balance needs an admin instead of a number. See Credits.

Audit log

Filter both the organization audit log and the platform-wide view (staff only) by date range, user, resource type and source, and export the filtered set to CSV, up to 50,000 rows.
Every entry now also records the device type, browser and OS parsed from the client, the country (and city and region, where Cloudflare provides them), and the source the request came from: web console, API token, MCP server, the Slack or Teams bot, or an internal or scheduled job. See Audit Logs.

Alerts

Alert rules can be edited in place rather than deleted and recreated, and editing re-evaluates the rule immediately against current data.
Send test sends a real notification through the rule’s configured channels (email, Slack, or both) without waiting for it to actually trigger, so you can confirm delivery while setting one up.
Alert notifications, including the daily digest and credit alerts, now render in one consistent branded Slack card layout.

Two factor authentication

2FA is now required for every organization, not an admin-configurable setting. Anyone signing in with a password, including choosing an organization, accepting an invitation, or verifying an email address, is asked for an authenticator code or a passkey, or walked through setting one up.
Signing in through SAML SSO is exempt, since your identity provider handles multi-factor authentication. A SAML account that tries to sign in with a password instead, and has no factor of its own enrolled, is sent to SSO sign-in rather than left stuck. See Security.

Team

Removing someone from your organization is now permanent (it deletes their membership, rather than only deactivating it), and separate from deactivating them. If the person owns anything, budgets, alert rules, reports, saved views, connected accounts and the rest, you choose a named successor before the removal completes; deactivating still leaves items in place, reachable by admins. See Team Management.

Notifications

Cloud sync now sends one summary email per organization per sync run, instead of one per connected account.
A notification that fails to send on one channel, such as a credit or budget alert over email or Slack, is retried automatically on the next check without resending the channel that already went through.
The three onboarding emails carry a one-click unsubscribe link. Unsubscribing stops only the onboarding series; verification, password reset, security notices and alerts you’ve configured are never affected.

Kubernetes

The console now shows the outbound host a private cluster’s collector sends to, ingest.xplorr.io:443, in the connect-cluster steps and as a tooltip on Waiting for collector, so you have it on hand before opening the collector dialog if your cluster sits behind an egress allowlist.
The public Helm chart (xplorr 0.1.0) now checks for an OpenCost already running in the cluster before installing its own, and refuses the install with the exact flags to point it at the existing one instead of installing a duplicate.

September 23, 2026

Security fixes

Closed a gap where cloud-sync, optimization and reporting accepted the token issued between your password and your second factor. A password alone could no longer reach those services once this shipped, even with 2FA required.
Refresh tokens can no longer be used in place of an access token.
A viewer’s account access is capped at read everywhere, not only in the console. Removing a cloud account or replacing its credentials is admin only; the person who connected it can still rename and sync it.
Passkey sign in is fixed. It had started failing to find the right credential during verification.
Verifying your email or accepting an invitation now goes through the same second factor check as a normal sign in, including on a bookmarked or reopened link.
Two ways a customer admin could reach Xplorr staff access through SSO are closed.

Permissions and platform staff

One permission table now decides what each role can do, read the same way by every service and the console. It is rolling out gradually: existing behavior does not change yet, and member permissions such as creating an org-wide alert take effect only once enforcement is switched on. Watch this page for that update.
Xplorr staff are now a distinct concept from your organization’s roles. Your roles stay viewer, member and admin.
Members can now share a cloud account they connected with the rest of the organization, read only.
Admins can reset a member’s second factor from the Team page, for people who belong to only one organization. Contact support for anyone in more than one.
Deactivating or demoting someone who owns budgets, alerts, reports, views or other items now offers to reassign them to someone else, instead of leaving them ownerless.
Sign-in history (successes and failures, with time, IP, browser and country) is now recorded and kept for 90 days.

SSO

Verify domain ownership for SAML SSO with a DNS TXT record. A domain does nothing for sign in until it is verified: no domain based routing, no SAML sign in, no auto-provisioning and no “require SSO for all” until then. See Team Management and Security.
Xplorr staff accounts can no longer sign in through SSO.

Recommendations

Enterprise organizations can send a recommendation for approval directly from the Recommendations page with a Request approval button, without leaving the console.

Email

Outbound mail now sends from mail.xplorr.io.

September 22, 2026

One account across organisations

One email address can belong to more than one organisation. Signing in now asks which one rather than picking for you, and your role, your account access and your second factor prompt all come from the organisation you choose.
Accepting an invitation leaves an account you can sign in to again afterwards. Previously the invitation signed you in once and the account could not be used after that.

One second factor, not one per organisation

An authenticator app or passkey is set up once and covers every organisation you belong to. It used to be stored per membership, so somebody in three organisations enrolled three times and an administrator resetting it fixed only one.
Turning it off, or having it reset, now applies everywhere.

Profile and organisation

Upload a profile photo, and a logo for your organisation. PNG, JPEG, GIF or WebP up to 1MB. The logo appears beside the organisation name in the sidebar.
Plans are named as they are sold: Starter, Growth and Enterprise.

Fixed

Saving settings cleared the organisation’s Slack webhook and reported success. Every Slack alert stopped until it was entered again.
The organisation-wide Slack webhook was readable by any member. It is administrators only.
Closing an organisation always failed.
The SSO panel reported auto-provisioning off and “required for all: no” regardless of what was configured, and neither switch could be changed without setting SSO up from scratch.
A profile with no last name could not be saved.
An organisation could be left with no administrator through the API.

September 22, 2026 — Kubernetes

Connect a Kubernetes cluster with one command

A public Helm chart at charts.xplorr.io installs OpenCost and the Xplorr collector together, so connecting a private cluster is one helm upgrade --install rather than installing OpenCost, configuring it and then applying a manifest.
The chart gives OpenCost its CLUSTER_ID from the cluster name you pass it, so the two cannot name different clusters. A cluster whose OpenCost still reported default-cluster used to look like it was working while filing its cost under another cluster’s name, and nothing said so.
The console shows the command with your ingest token already in it, plus the one that runs the collector immediately instead of waiting for the hour.
Already running OpenCost? --set installOpenCost=false and point the chart at it. The YAML manifest is still there for anyone who would rather use kubectl.
Source, values and what actually runs in your cluster: Xplorrio/charts.

Fixed

Generating an ingest token or downloading the collector manifest returned a server error. Connecting a private cluster could not be completed.

September 21, 2026

Two factor authentication, with passkeys

2FA can be required across the platform, and a passkey satisfies it: the fingerprint, face or PIN you already unlock your device with, or an authenticator app if you prefer a code.
Anyone who has to set one up is walked through it at sign in rather than being locked out.
An administrator can reset a second factor for someone who has lost their device.

Members can own what they connect

A member can connect an AWS, Azure or GCP account and manage it, including granting other people access to it, without being made an organisation admin.
A member can connect a Kubernetes cluster. Deleting one still needs an admin.

Cost detail

Network cost drivers open in place to show the resources behind them: which peering, which gateway, which DNS zone. Where a provider does not collect per-resource network costs, the page says so rather than showing an empty list.
Data transferred shows a dash where the meter does not count bytes, instead of 0 B. A VPN gateway bills by the hour.
A newly connected account backfills 90 days instead of 30, so last month is complete on the day you connect it.
Allocation coverage compares cluster cost against the cloud bill for the days the cluster actually reported, rather than the whole window, and the table says how many days that is.

Fixes

Connecting an Azure subscription verifies the subscription rather than running a cost query, so it completes in a second instead of over a minute, and names the missing permission when one is missing.
Removing someone from an organisation now removes them from the team list, and inviting the same address again restores their account.
Alert rules can be scoped to an account from the form, so a member can create one.
The Connect cluster form no longer lets a browser autofill the cluster name with the page URL, and explains that OpenCost’s CLUSTER_ID is a separate thing from the name you give a cluster here.

September 20, 2026

Cost collection moved onto a job queue

Per-account syncs now run on a queue rather than one after another in a single loop, so accounts sync concurrently instead of at roughly a minute each in sequence.
A sync that fails is retried with backoff, and one that fails every attempt is kept in a dead letter queue with the account, the window it covered, how many attempts it made and why it failed.
Data Health gained a Sync queue panel showing what is waiting, what is running and what failed, with a Retry button on each failure.
Sync now on any account queues a sync immediately and returns a job id you can follow, instead of starting work in the background with no way to see it.
A failure before a sync begins, such as an account removed between queueing and running, is now visible. Previously it wrote no record anywhere.

GitHub as a cost source

Connect a GitHub organisation to bring Actions minutes, storage and Copilot seats in beside your cloud spend.
Billing is mapped to the same cost model as every other source, so GitHub appears in allocation, budgets and reports rather than in a page of its own.

Plan limits

Connected accounts, seats, API tokens, budgets, alert rules and scheduled reports each have a published limit per plan, shown on the pricing page.
Reaching a limit is explained in the console with what the limit is and which plan raises it, rather than a refusal without a reason.
An SSO login is never blocked by a seat limit.

September 18, 2026

Kubernetes cost dashboard

Cluster costs with efficiency, per-node and per-volume breakdowns, and commitment coverage.
Push-based ingest for private clusters, so a cluster that cannot be reached from outside can still report its costs over outbound HTTPS.

September 16, 2026

Transactional email redesigned

Every email Xplorr sends now shares one design system, so an alert, a digest and an invitation look like the same product.

Local development stack

A synthetic demo dataset and a Docker Compose stack, so the product can be run end to end without connecting a real cloud account.

September 15, 2026

FinOps framework coverage

Gaps closed across billing, commitments, allocation and invoicing to line up with the FinOps Foundation’s framework.

AI provider choice

AI features can run on OpenAI, OpenRouter or AWS Bedrock.

Console accuracy and interface pass

Corrected the 30-day forecast, month-over-month comparison and savings coverage totals, each of which could report a figure that did not match the underlying data.
Charts and forecasts no longer treat the incomplete current day as a full day.
Azure inventory status and network costs now collect correctly, and cost tags are ingested so Cost Allocation reflects real tagging rather than reporting everything as untagged.
Consistent number, currency and date formatting throughout, official provider and service icons, and a dark mode pass.

March 18, 2026

Network cost visibility

Added dedicated Network Costs page with per-provider breakdown of egress, inter-AZ, cross-region, CDN, VPN, and VPC peering charges.
NAT Gateway cost tracking with daily processing volume and hourly charges shown separately.
Top network cost drivers ranking to identify the biggest optimization opportunities.

Commitment planner improvements

The RI/SP commitment planner now shows 1-year vs 3-year side-by-side comparison for each candidate workload.
Added usage consistency score to help assess commitment risk.
Planner accounts for existing commitments and their expiration dates.

March 17, 2026

Tagging policy enforcement

New tagging policies feature: define required tags, allowed values, and naming conventions.
Tagging violations page shows all non-compliant resources with links to fix them.
Untagged cost trend chart tracks tagging coverage improvements over time.

Report scheduling

Scheduled reports now support daily, weekly, and monthly delivery.
Added external recipients — send reports to email addresses that don’t have Xplorr accounts.
Chargeback report format with shared cost allocation rules.

March 15, 2026

Seasonal forecast model

Forecasting now uses a day-of-week seasonal model when significant weekly patterns are detected.
Forecast accuracy history shows projected vs actual EOM spend for the past 6 months.
Projected budget overrun alerts fire earlier — now based on forecast data rather than linear extrapolation.

Security updates

WebAuthn/passkey support for passwordless authentication.
Progressive account lockout (1 min → 5 min → 15 min → 1 hour).
Rate limiting headers now included in all API responses.
Audit log retention extended to 90 days.

Bug fixes

Fixed an issue where Azure cost data for accounts with more than 50 subscriptions would timeout during sync.
Fixed recommendation savings estimates not accounting for existing reserved instance discounts.
Fixed weekly digest emails showing UTC timestamps instead of the user’s configured timezone.

March 10, 2026

AI-generated remediation scripts

Recommendations now include CLI commands, Terraform snippets, and manual steps tailored to each resource.
Scripts are provider-specific and account for your current configuration.
“Show all remediation options” expands to display all three formats.

Notification improvements

Per-user notification preferences: toggle each notification type independently.
RI/SP expiration alerts at 30, 14, and 7 days before expiry.
Slack notifications now include inline cost data and a direct link to the relevant console page.

March 5, 2026

Coverage gap analysis

New Coverage Gaps tab identifies instance families running on-demand with consistent usage patterns.
Each gap shows average daily hours, consistency score, and potential monthly savings.
Gaps are ranked by savings to prioritize the highest-value commitments.

Team management

Per-account access control: restrict which cloud accounts a user can see.
Bulk user invites with comma-separated email addresses.
Org-wide 2FA enforcement toggle for Admins and Super Admins.

February 2026 and earlier

For updates prior to March 2026, see the full changelog on the Xplorr website.