Overview

Network Topology Explorer (Open Beta)

Network Topology Explorer is now available to every partner as an open beta. It builds a live, visual map of how devices in a customer's network relate to each other — subnets, gateways, switch ports, and routes — using data your inspectors already collect, so technicians can trace a connectivity issue visually instead of cross-referencing device records by hand. The explorer sits at the top of Asset Inventory > Devices as a collapsible card with a BETA badge, collapsed by default.

What's New

• Subnet overview that shows each subnet as a bubble sized by device count, with drill-in per subnet or a full graph across the Environment
• Five layouts — Tiered (default), Grouped, Force, Radial, and Grid
• Find node to search for a device, center the map on it, and open its detail panel
• Four analysis lenses: L2 (Switched), L3 (Routed), Exposure (devices with a public IP), and Freshness (devices not observed in more than 30 days)
• Path Finder to highlight the shortest connection path between a source and a target, with hop count
• Focus view — double-click a device to load just that device and its direct neighbors
• A per-device detail panel with filter icons that filter the Devices table below
• Unmanaged device detection that shows devices seen by a supported network inspector, but not yet in inventory, as dashed "ghost" nodes
• Fullscreen mode, plus two new optional Devices table columns: Default Gateway and Primary Subnet CIDR

What You Need to Know

Any device with a subnet or default gateway appears on the map, including Network Discovery results. Device-to-device connection data — switch ports, VLANs, wireless clients, and routed paths — comes from the SonicWall, HP ProCurve, Cisco Meraki, and Fortinet FortiGate Inspectors. There's no historical backfill: the map populates as each launchpoint completes its next inspection after this release, so you may see a partial map at first. No new licensing, integration, or permissions are required.

To learn more about this feature, please review our documentation.

Asset Inventory: 12 New Identity and Device Sources

Asset Inventory now pulls identity and device data from 12 additional inspectors, giving you a more complete, unified view of the accounts and assets you manage. Records that were previously visible only inside each inspector's own System view now appear in Asset Inventory, where you can filter, search, and report on them alongside everything else.

New Identity Sources (Asset Inventory > Identities & Accounts)

• Anthropic Console — organization members and pending invites
• Windows Workstation — local user accounts
• AWS — IAM users, including MFA status
• Azure — Azure users
• SentinelOne — console users, including MFA status
• Synology NAS — NAS user accounts
• BitLyft — internal console users, including privileged access status
• Microsoft 365 Education — teachers and students from school classes (requires School Data Sync). Teachers are marked Billable and students are marked Non-Billable.

New Device Sources (Asset Inventory > Devices)

• Fortinet FortiManager — FortiGate firewalls managed by FortiManager
• SonicWall Capture Client — endpoints protected by Capture Client
• Barracuda CloudGen Firewall — Barracuda firewall appliances
• Datto Networking — access points and switches

UniFi Cloud devices now also show device status, first seen date, and site/location.

Benefits

• Run access reviews across more tools, with AWS, Azure, SentinelOne, BitLyft, and Anthropic Console accounts sitting beside your directory identities.
• Surface local Windows Workstation and Synology NAS accounts that directory-based reviews often miss.
• Build a more complete network inventory with firewalls, access points, and switches from FortiManager, Barracuda, and Datto Networking.
• Track who has access to your clients' AI tooling with Anthropic Console members and pending invites.

When more than one inspector reports the same device, the sources reconcile into a single record that lists each related system. No setup is required — new records appear automatically after each inspector's next successful inspection, using existing credentials.

Asset Inventory: Richer Device Data from RMM and Endpoint Management Inspectors

Asset Inventory now shows much more of the device data Liongard already collects from your RMM and endpoint management inspectors, so you get a complete device record without deploying the Liongard Endpoint agent.

What's New

• Added hardware identity, operating system version, status, site, first and last seen, warranty and lifecycle dates, and physical or virtual details for Datto RMM, ConnectWise Automate, ConnectWise Asio, NinjaOne, N-able RMM, N-able N-central, Kaseya VSA, Syncro, and Addigy (exact fields vary by inspector)
• Network devices and printers managed in NinjaOne and ConnectWise Automate now appear with the correct category and type instead of as computers
• Last Seen now reflects the last activity reported by the integration, not the latest inspection time, across 12 inspectors: Active Directory, Microsoft 365, Duo Security, Sophos Central, Watchman Monitoring, Huntress, JumpCloud, Axcient x360Recover, StorageCraft SPX, 3CX, NinjaRMM, and ConnectWise Automate

With serial number and MAC address available, devices reported by both an RMM and other inspectors match into a single record more reliably, and warranty, purchase, and replacement dates support hardware refresh planning. No setup is required — details update automatically after each inspector's next successful inspection.

Asset Inventory: Automatic Managed Status, Billable Device Count, and Clearer States

Asset Inventory now decides automatically which devices are Managed, based on the inspector that discovered them, and uses that status to count billable devices. The Asset Inventory states also have new names that say exactly what they mean.

What's New

• Devices are marked Managed or Unmanaged automatically. Endpoint and network inspectors mark devices Managed by default; EDR and RMM inspectors do so when their data shows an agent is installed; Microsoft 365 does so when the device is Intune-managed. Being listed only in a source such as Active Directory, Duo, or 3CX doesn't make a device Managed.
• A device moves to Unmanaged when it hasn't been seen for 45 days, and returns to Managed when it's seen again
• A device counts as billable when it's Managed, has been seen within the last 45 days, hasn't been deleted, and isn't a printer, smartphone, or tablet
• Turn off Auto update for a device's Managed setting to override the automatic value; billing then follows the value you set
• Discovery is now Discovered, Inventory is now Managed, and Archive is now Unmanaged, for both devices and identities, with actions renamed to Move to Managed and Move to Unmanaged
• Devices and identities in the Unmanaged state can now be deleted. If a later inspection finds one again, it comes back as newly Discovered.

What You Need to Know

No action is required. If you have saved filters, reports, or internal documentation that reference the old state names, update them to the new names.

Liongard Roar Assistant: Expanded AI Model Selection and Allowed AI Model Providers

The Roar Assistant model picker now offers a broader set of AI models, and admins get a new control to decide which AI model providers their company can use at all.

What's New

• The model picker, next to the Send button in the Roar Assistant chat, now includes models from Anthropic, OpenAI, xAI, Moonshot AI, and Z.ai, with Auto remaining the recommended option
• Each model shows its AI Insight Credit rate in the picker, and the first time you choose a model other than Auto, a billing acknowledgement confirms its rate
• The picker lists only models that can run in your instance's region, so not every instance shows every model
• A new Allowed AI model providers multi-select in Company Settings > AI INSIGHTS lets admins choose which providers' models appear in the Default AI model list and the chat picker, and which providers Liongard routes AI requests to. All providers are selected by default.

AI Insight Credit Rates

Rates are per model, per request. A 2× model uses twice the allowance of a 1× model for the same request, and a 4× model uses four times as much.

• 1× — Claude Haiku 4.5, Claude Sonnet 5, Claude Sonnet 4.6, GLM 5, Grok 4.7, Kimi K3, and Kimi K2.5
• 2× — Claude Opus 5.5 and Claude Opus 5
• 4× — GPT-6.1 Sol

Auto bills at the rate of whichever model served the request. Models with a rate above 1× are never selected by Auto, so using one is always an explicit choice by a user or admin.

What You Need to Know

The provider control applies company-wide across every AI feature, including Roar Assistant chat, liongard_query, agents and recipes, and background AI tasks such as summarization, and it never falls back to a cleared provider. If an admin clears the provider of the current Default AI model, the default resets to Auto. To pin everyone to one specific model, set the Default AI model and turn off Allow user overrides. Higher-rate models use your AI Insight Credits faster, which can increase overage if Pay-as-you-Go Billing is enabled.

To learn more about this feature, please review our documentation.

Liongard MCP: OAuth Sign-In

The Liongard MCP server now supports OAuth 2.0 sign-in. Instead of creating, copying, and pasting a static Bearer Token into your AI tools, you connect an MCP client to Liongard with a browser sign-in and one-click approval.

What's New

• Add the Liongard MCP endpoint (https://[server].app.liongard.com/api/mcp) to your MCP client, choose Authenticate, sign in, review the client name and requested scopes, and select Approve
• Consent runs on your existing Liongard login, so SAML SSO and MFA requirements are honored with no extra setup
• Tokens carry the user's identity, tenant, and environment scope, so MCP activity is attributable to a user rather than a shared token
• Access is limited to the mcp.read and mcp.write scopes and the user's existing environment scope
• Short-lived access tokens refresh automatically and can be revoked at any time
• Supported clients include Claude Code, OpenAI Codex, and Zed, with more to follow

OAuth is enabled per service provider, and existing Bearer Token connections continue to work.

To learn more about this feature, please review our documentation.

Liongard Agent: Self-Update, Release Trains, and Update Windows

Starting with Agent 5.5.0, update checking moves into the agent itself. Instead of checking once a night, agents check for updates on every heartbeat — about every 3 minutes — so a laptop that was off overnight picks up its pending update shortly after it comes back online.

What's New

• Heartbeat-driven self-update on Windows, macOS (including Apple Silicon), and Linux (including ARM64)
• Release trains you can choose at the organization level or per agent: Early Access (EAP), Stable, and Long-Term Support (LTS). Agents with no release train set follow LTS.
• Daily update windows (for example, 02:00 to 04:00), set organization-wide or per agent and enforced on each agent's own local clock
• A Failed Updates tile and an Update Health column on the Agents list, plus a failure alert on the agent page with the stage, error, attempts, next retry, and installer log tail
• Built-in retries — up to three attempts about 10 minutes apart, then once a day
• Downloads verified against a SHA-256 checksum from agents.static.liongard.com before anything runs
• An updated download menu with per-platform options, including new macOS Apple Silicon and Linux ARM64 installers

What You Need to Know

Updates can now land during the business day. If timing matters, set an update window under Company Settings > Agents > ORGANIZATION UPDATE SCHEDULE (windows can't cross midnight). Release train and update window settings apply only to agents on 5.5.0 or later. An update waits for running inspections, and the machine doesn't reboot. The Auto Update setting (Enabled / One Time Update / Disabled) is unchanged — set it to Disabled to hold an agent on its current version.

To learn more about these new agent features, please review all our documentation, Agent Auto-updating, Agent Release Trains, Agent Update Windows.

Liongard Agent: Duplicate Agent Detection and Auto Resolve

When partners clone VMs or deploy machines from a golden image, the clones can end up sharing one agent identity. Liongard then treats them as a single agent — the record flips between hostnames, IP addresses, and versions, Inspectors run on whichever machine checked in last, and the other machines never show up. Duplicate Agent Detection and Auto Resolve lets you find and fix these agents yourself in a few clicks, with no support ticket, no wrapper script, and no changes to the operating system's machine ID.

What's New

• Starting with Agent 5.5.0, every fresh install gets its own Liongard-managed identity instead of reusing the operating system's machine ID, so machines cloned before the agent was installed register as separate agents automatically
• Liongard watches agent check-ins and flags any agent that more than one machine is reporting as
• A Duplicate IDs card and a Duplicates column on Admin > Agents show flagged agents, with Suspected, Resolving, and None statuses
• A warning banner on the agent page, with a Duplicate Identity window that shows the hostnames and IP addresses seen and what triggered detection
• Auto Resolve tells every machine sharing the identity, except one, to re-enroll as its own new agent in the same Environment, named {Agent Name} (1), {Agent Name} (2), and so on. The machine left behind keeps the original record, its history, and its Inspectors.
• A Golden Image toggle that makes every machine cloned from a template re-enroll as its own agent automatically, with no detection wait
• New identity reset options for image prep: --reset-identity on Linux and RESETDEVICEGUID=1 on Windows
• Six new activity log events for detection, Auto Resolve, re-enrollment, and Golden Image changes

What You Need to Know

Detection and Auto Resolve work only for self-managed agents on 5.5.0 or later, and detection needs the machines sharing an identity to be online at the same time, typically for 10 to 20 minutes. Liongard doesn't email or alert you when it flags a duplicate — check the Duplicate IDs card and the activity log. Auto Resolve can't tell which machine was the original, so after a round completes, check where Inspectors landed and move them if needed. New agents start with no Inspectors and follow your organization's default release train and update schedule. Only arm Auto Resolve on an agent that really has more than one machine behind it; there is no undo.

To learn more about this feature, please review our documentation.

NinjaOne Inspector Enhancement: Complete Data Collection and Expanded Device Details

The NinjaOne Inspector now collects every page of data from the NinjaOne API and adds eight new data sets, giving you a complete, accurate view of each NinjaOne-managed environment.

What's New

• Complete pagination. The inspector now follows pagination on every endpoint. Previously, only the first page of results was collected, so large tenants could under-report counts in metrics and miss child inspectors for organizations beyond the first page.
• Eight new data sets: Operating Systems, Processors, Network Interfaces, RAID Controllers, RAID Drives, Software Patch Install History, Policy Overrides, and Backup Usage.
• Existing data, metrics, and views are unchanged — the new data sets are purely additive.

What You Need to Know

The new data sets are available right away for building your own metrics and alerts. No new out-of-the-box metrics or views ship with this update. Backup Usage data appears only for tenants licensed for NinjaOne Backup. No setup changes are required; the new data appears after each launchpoint's next inspection.

Huntress Inspector Enhancement: Incidents, Signals, Escalations, and Expanded Coverage

The Huntress Inspector now goes well beyond account, organization, and agent data. It collects the security and operational data Huntress has added to its platform, so you can see incidents, items needing attention, and coverage gaps for each customer without logging into the Huntress portal.

What's New

• Incident reports for each account
• Signals — SOC-level behavioral detections, from suspicious command-line activity to known malware
• Escalations that need partner or admin attention, such as false-positive reviews and login-anomaly confirmations, with severities
• Platform actions — operational health and misconfiguration warnings, such as a broken PSA integration, disabled Microsoft Defender, failing Microsoft 365 authentication, a log source that's down, or low disk space
• External ports — exposed ports and services per account from Huntress External Recon
• Identities and memberships, adding user visibility the inspector didn't have before
• Unwanted access rules and known VPNs from Huntress identity threat detection and response (ITDR)
• Remote access connections, for visibility into RMM and remote access tools in use

What You Need to Know

All existing account, organization, and agent data, metrics, and alert rules are preserved. The inspector remains read-only — it doesn't isolate agents, resolve incidents or escalations, or change any Huntress configuration. No setup changes are required; the new data appears after each launchpoint's next inspection.

SonicWall NSM Inspector (Beta)

SonicWall Network Security Manager (NSM) is SonicWall's cloud console for centrally managing a fleet of SonicWall firewalls. You can now use the SonicWall NSM Inspector (Beta) to connect to a customer's NSM tenant, discover every firewall it manages, and bring firmware posture, configuration backups, change history, alerts, and security-service status into Liongard — so an offline firewall, a missing backup, or a disabled security service surfaces without logging into NSM.

What you can monitor

• Managed firewall inventory by model and SonicOS version
• Offline firewalls and firewalls that haven't checked in for 24 hours
• NSM admin users, including accounts without MFA and stale accounts
• Firmware version and whether newer firmware is available, per firewall
• Configuration backup recency and recent configuration commits
• Active and critical alerts
• Gateway Anti-Virus, Intrusion Prevention (IPS), and Content Filtering status

Benefits

• Catch offline or silent firewalls before a customer reports an outage.
• Confirm every firewall has a recent configuration backup so a recovery is always possible.
• Flag outdated firmware and disabled security services that weaken cyber resilience.
• Run side by side with the existing SonicWall and SonicWall Capture Client Inspectors — the NSM Inspector complements them, it doesn't replace them.

Beta: This inspector is currently available as a Beta feature. Functionality may evolve as we continue to enhance the integration based on partner feedback.

Please review our documentation to learn more about this inspector.

Palo Alto Prisma Browser Inspector (Beta)

Palo Alto Prisma Browser is a secure enterprise browser that governs which websites and AI applications users can reach and which data-handling controls apply. You can now use the Palo Alto Prisma Browser Inspector (Beta) to discover every customer tenant your MSP manages through the Prisma Browser Partner Provisioning API, and track each tenant's status, SSO, users, content filtering, and data-loss-prevention (DLP) controls — so you can spot coverage gaps and policy drift without opening the Prisma Browser console for each customer.

What you can monitor

• Tenant inventory and lifecycle status (active, under review, suspended)
• SSO configuration status
• Provisioned users and admin users
• Subscription type and trial end date
• Content-filter categories and individually blocked websites
• DLP data control rules, including rules with a protection disabled

Benefits

• Validate that every customer's Prisma Browser configuration matches your security baseline.
• Detect policy drift — removed content-filter categories, removed or weakened DLP rules, and disabled SSO — as change events.
• Catch tenants moving to under review or suspended, or trials nearing their end, before protection lapses.
• Keep an evidence trail of browser and AI-governance configuration to prove value in QBRs and security reviews.

Beta: This inspector is currently available as a Beta feature. Functionality may evolve as we continue to enhance the integration based on partner feedback.

Please review our documentation to learn more about this inspector.

Blackpoint Cyber Inspector (Beta)

Blackpoint Cyber's CompassOne platform delivers Managed Detection and Response (MDR) across endpoint and cloud environments. You can now use the Blackpoint Cyber Inspector (Beta) to connect to CompassOne with a single partner-level API key, discover each managed customer tenant, and bring security posture, detections, assets, vulnerabilities, and dark-web credential exposures into Liongard — without logging into CompassOne for each customer.

What you can monitor

• CompassOne security posture score, status, and maturity level, with the deductions behind the score
• Detections and their open or resolved status
• Asset inventory across devices, users, software, sources, and surveys
• Vulnerability findings by severity, and vulnerability scan status
• Dark-web credential exposures

Benefits

• Track every customer's security posture score over time and catch drops as change events.
• Surface new critical vulnerabilities and failed vulnerability scans.
• Flag new dark-web credential exposures before those credentials are used.
• Prove the value of your MDR service in QBRs with posture, detection, and exposure history.

Beta: This inspector is currently available as a Beta feature. Functionality may evolve as we continue to enhance the integration based on partner feedback.

Please review our documentation to learn more about this inspector.

Axcient x360Cloud Inspector (Beta)

Axcient x360Cloud is a cloud-to-cloud backup product for Microsoft 365 and Google Workspace. You can now use the Axcient x360Cloud Inspector (Beta) to connect to your x360Portal account with a single API key, discover every customer organization you protect, and bring backup health, protected users, and shared resources into Liongard — so a failing backup or a user silently dropped from protection shows up without logging into the x360Portal.

What you can monitor

• Latest backup status and backed-up user, SharePoint site, and shared drive counts
• Protected and billable user counts, and per-service protection settings
• Every protected user and shared mailbox, with protection flag, billing status, and latest backup status
• Every SharePoint site and Google Shared Drive, with protection flag and latest backup status
• Available seats on your not-for-resale (internal) organization

Benefits

• Spot a customer whose Microsoft 365 or Google Workspace backup has started failing before they notice data loss.
• Confirm every protected user and shared resource is still selected for protection, and catch silent de-provisioning.
• See cloud backup health in the same place you track the rest of each customer's configuration.

x360Cloud is a separate product from Axcient x360Recover, which has its own Liongard Inspector.

Beta: This inspector is currently available as a Beta feature. Functionality may evolve as we continue to enhance the integration based on partner feedback.

Please review our documentation to learn more about this inspector.

Bitwarden Inspector (Beta)

Bitwarden is a password and secrets manager that organizations use to store and share credentials securely. You can now use the Bitwarden Inspector (Beta) to connect to a Bitwarden organization through the Bitwarden Public API and track its security policies, members, two-step login (2FA) status, groups, and collections — so you can see who holds Owner and Admin rights, who's missing 2FA, and whether vault security policies stay enforced, without logging into the Admin Console.

What you can monitor

• Require Two-Step Login, Master Password Requirements, Single Organization, and Require Single Sign-On policy enforcement
• Members, with role, account status, and two-step login status
• Owner, Admin, and revoked member counts
• Groups and collections

Benefits

• Catch a disabled two-step login, master password, or SSO policy as a change event.
• Track new Owners and Admins so privileged vault access never changes unnoticed.
• Identify members without two-step login and close 2FA gaps across your customer base.
• Populate a Bitwarden Organization asset with member counts, 2FA gaps, and policy-enforcement status.

Beta: This inspector is currently available as a Beta feature. Functionality may evolve as we continue to enhance the integration based on partner feedback.

Please review our documentation to learn more about this inspector.

Cisco Business Dashboard Inspector (Beta)

Cisco Business Dashboard (CBD) is a self-hosted virtual appliance that MSPs use to manage Cisco 100 to 500 Series routers, switches, and wireless access points. You can now use the Cisco Business Dashboard Inspector (Beta) to connect to a Dashboard instance, discover every network (site) it manages, and track device inventory, firmware currency, end-of-life (EOX) notices, warranty coverage, and availability — without logging into the Dashboard.

What you can monitor

• Organizations, networks (sites), and device groups
• Each network's online state and unacknowledged alert and warning counts
• Device inventory per site, including model, serial number, IP and MAC address
• Current and latest available firmware per device
• EOX and last-date-of-support dates, warranty and coverage end dates
• Online, offline, and blocked devices

Benefits

• See which devices on each site are behind the latest firmware, and catch unplanned upgrades or rollbacks.
• Build replacement and renewal lists from hardware with an EOX notice or no active coverage.
• Know when a site, or a device on it, drops off — and catch devices that appear on or disappear from a managed network.
• Surface sites with Dashboard alerts nobody has acknowledged.

This inspector is separate from the Cisco SBS Switch Inspector, which connects to a single switch over SSH. The two can run side by side.

Beta: This inspector is currently available as a Beta feature. Functionality may evolve as we continue to enhance the integration based on partner feedback.

Please review our documentation to learn more about this inspector.

---

Minor Updates and Bug Fixes

Platform and Gen AI Minor Updates and Bug Fixes

• You can now turn off Auto update on an Asset Inventory value you've edited so the next inspection doesn't overwrite it. An auto-update toggle replaces the field lock icon, and the control now covers dropdown and toggle fields (including bulk edits) on Device, Identity, and Domain summaries, as well as lifecycle dates.
• Added a comment field to the bulk alert close action. The comment is applied to every closed alert's history.
• Fixed an issue where bulk-closing alerts showed a success message but the alert status did not update in the UI.
• Inspectors whose Liongard Agent was deleted now appear on the Unmanaged Inspectors tab so you can review and clean them up in bulk, and the tab's count is now correct.
• Added Managed and Unmanaged classification for domains, and Managed and billable status logic for identities in Asset Inventory.
• Fixed an issue where phones were categorized as billable compute devices.
• Fixed an issue where Dark Web Monitoring identities were ingested with MFA set to "No," inflating MFA-disabled counts.
• Fixed an issue where distinct Windows workstations were merged into a single device profile because they shared an OEM placeholder serial number.
• Fixed an issue where parent inspectors ingested child devices from across the fleet into Asset Inventory, causing duplicates across Environments, and where devices discovered by parent inspectors remained after their data was removed.
• Fixed an issue where the Asset Inventory export showed outdated Inventory State values compared to the UI.
• Added missing inspector-specific columns to the Asset Inventory Devices table, added missing inspectors to the Related Systems filter, and fixed the filter's ordering for Devices and Identities.
• Fixed an issue where the device create API returned a server error instead of a validation error when Alias was omitted, and where manual device creation accepted a whitespace-only Custom Type.
• Fixed an issue where the required Label/Name field accepted whitespace only and still allowed credentials to be generated.
• Fixed an error in the v1 Timeline API.
• Fixed an issue where long metric queries could fail.
• Fixed an issue where the Expand button did not work in the Rules Currently Enabled pop-up on the Environments tab.
• Fixed an issue where Network Topology Explorer subnet overview nodes overlapped and hid labels.
• Fixed an issue where values in the Home page open alerts table were not fully visible.
• Fixed an issue where the All managed environments group included the Internal Environment.
• Fixed an issue where the Identity and Device Inventory tabs were missing from the home dashboard after switching back to the legacy UI.
• Fixed an issue where a white highlight stayed on an Admin menu item after navigating in the new top bar.
• Fixed an issue where saved Quick Views did not keep their column selections, and table column widths now persist across page reloads.
• Fixed an issue where right-click Copy on a System page cell produced "[object Object]" instead of the cell value.
• Fixed an issue where the inspector quick-nav dropdown did not navigate correctly.
• Fixed an issue where the Actionable Alert silence window did not display the final silence option.
• Fixed an issue where line breaks in an Actionable Alert body showed as extra characters.
• The Documentation button now appears in the Actions column of the Admin > Inspectors table for all roles.
• Fixed unreadable note chips in the Inspector Details table and low-contrast AI Summary text in dark mode.
• Fixed an issue where the Show More button on the Environment column in AI Preferences opened and immediately closed.
• Fixed misplaced scoping chips in the Roar Assistant.
• Fixed an issue where the Create Report option inside a report template did not work.
• Fixed an issue where several Liongard reports failed to load and download, and where PDF export failed when report data was too large.
• Fixed an issue where deleted Environments appeared as "undefined" in Excel report downloads.
• Fixed misaligned action buttons in the report creation and report template modals.
• The in-app UI now uses AI Insight Credits in place of "AI Credits."
• In-app AI summaries for assets, Environments, and systems are now metered at 1 AI Insight Credit per completed summary.
• Fixed an issue where direct-MCP liongard_query calls were metered in the 1-per-100 tool-call bucket instead of at 1 credit per query.
• Added a billing acknowledgement for MCP Private Access Tokens and a Created By column on the Access Tokens list.
• The MCP usage screen now shows credit consumption by user account and by token.
• Added persona-based suggested prompts to the Roar Assistant.
• The Liongard MCP server now supports MCP protocol version 2026-07-28 and a CSV response format.
• Added OAuth connect, discovery, and status for registry MCP servers in the MCP admin.
• Improved MCP OAuth compatibility with Cursor, Microsoft Copilot Studio, OpenAI Codex, and CLI clients that use loopback redirect URIs.
• Fixed several Roar Assistant accuracy issues, including internal IDs returned instead of friendly names, incorrect inspector information, an incorrect firewall count, errors during device lookups, SSL certificate expiration pulled from the Domain inspector instead of the SSL inspector, and inaccurate or irrelevant responses.
• Fixed an issue where the Liongard Agent did not display its internal IP address.
• Fixed an issue where the MAC address showed blank in Liongard Agent data for Windows Server and Workstation.
• Fixed an issue where Windows and Linux agent installation failed when the proxy authentication password contained an "@" character.
• Fixed an issue where the Windows agent reported version 1.0.0.0 instead of its real version.
• Fixed an issue with the Windows agent MSI installer's agent registration step when saving agent credentials.
• Added a description field to the macOS agent installer.
• Fixed an issue where macOS and Linux inspectors failed after updating to Agent 5.5.0.
• Updated dependencies to address security vulnerabilities in the Liongard Agent and in the pfSense, Active Directory, N-able N-central, Palo Alto, N-able RMM, and Sophos XG Inspectors.

Integration and Inspector Minor Updates and Bug Fixes

• Addigy
  • Fixed an issue where discovered devices did not populate into Asset Inventory.
• Auvik
  • Fixed an error when reassigning a child inspector.
• Autotask
  • Fixed a regression when adding and deleting statuses on the ticketing page.
  • Fixed an issue where Generate Test Ticket failed silently and created no ticket.
• Cisco IOS
  • Fixed an issue where the configuration drift metric always returned true.
• ConnectWise
  • Fixed an issue where an alert stayed open in Liongard after its ConnectWise ticket auto-closed.
• ConnectWise Asio
  • Fixed an issue where inspections reported success despite a 401 authentication failure and empty data.
• Domotz
  • Fixed child inspectors failing with a 412 error.
• KnowBe4
  • Fixed Account Activity mapping in Asset Inventory.
• Lexful
  • Updated the inspector setup help text.
• LogicMonitor
  • Fixed inspections failing with an HTTP 400 error on the websites endpoint.
• Microsoft 365
  • The inspector now collects mailbox delegation permissions (Full Access, Send As, and Send on Behalf), so you can audit who can reach which mailboxes. Changes can trigger alerts.
  • CSP-managed tenants now show each subscription's billing cycle, term, end date, and auto-renew status, so renewal alerts can target Annual licenses and skip Monthly ones.
  • Renewed (superseded) subscriptions no longer trigger expiration alerts.
  • Added Password Hash Sync last sync time to the Directory Sync dataset.
  • Added new Intune access-control, Settings Catalog, and Entra password-policy and branding metrics, plus success counts on the Intune policy metric.
  • Fixed an issue where the Save button kept loading when creating a launchpoint without Microsoft sign-in.
• Microsoft 365 Education
  • Fixed SSO and a launchpoint failure with a 400 error.
• N-able N-central
  • Fixed a setup issue where the inspector could not retrieve a valid customer ID.
• Network Discovery Inspector
  • Fixed an issue where Windows devices were still misclassified as SonicWall devices.
• Palo Alto
  • Fixed incomplete Admin Users data that caused false Change to Admin Users alerts.
• SentinelOne
  • Fixed an issue where child inspectors discovered unrelated customer tenants.
• SonicWall
  • Fixed the Policy Mode misconfiguration alert.
• SQL Server Inspector
  • Fixed false positive alerts for privileged user changes.
• ThreatLocker
  • Fixed an issue where inspections failed with an HTTP 904 error when Configuration Manager is unlicensed.
• TLS/SSL
  • Fixed false positive Web Traffic Exposure to Sniffing and Lack of Website Validation alerts for parked domains.
• UniFi Cloud
  • Fixed child inspectors failing with a 404 error, a timeout when accessing site-level controllers through the UniFi connector, and a hard failure on HTTP 400 when collecting firewall zones.
  • Fixed missing data in the site-list response.
  • Non-owner hosts are now flagged during parent discovery.
• Windows Workstation
  • Fixed an issue where the Windows Workstation and Roar Inspectors reported slightly different MAC addresses.
• 3CX
  • Fixed an issue where the inspector failed with a persistent 400 error when querying the backups endpoint.
  • Fixed an issue where an API row count limit could cause incomplete data.

Visual Insights Minor Updates and Bug Fixes

Features

• Roar Designer
  • Roar Designer can now inspect rendered Query Widget previews — charts, tables, and metrics as they appear on screen — when reviewing or refining your work.
  • You can now upload, paste, or drop files into Roar Designer chat so it can refer back to them later in the conversation.
  • Opening Roar Designer without a specific session now resumes your most recent conversation if you used it within the last hour. New conversations still start blank.
  • Dataset provisioning is safer: proposals add columns by default, keep sync settings you didn't change, reject outdated changes, and show clearer catalog and sync status before and after you apply them.
  • Roar Designer can now inspect source view definitions and shows clearer incremental-sync evidence during catalog review.
• Dashboards and reporting
  • Dashboards now remember each user's last filter selection and restore it the next time they open the dashboard, including in new tabs. Share Links, embeds, and view-as don't overwrite a user's saved filters.
  • Widget drilldowns can now be downloaded as Excel files, keeping configured column labels and order, with a CSV fallback for very large exports.
  • Dashboard table and matrix widgets now support pagination and a full-screen expand view, with J/K keyboard shortcuts while the pointer is over the widget. Query Widget tables paginate by default and can expand to preview the full result.
  • Gauge, Metric Display, bar, and line Query Widgets can now play a sound when a threshold is crossed, with mute and snooze controls on dashboards and Share Links. Thresholds also get directional colors, range legends, and clearer comparisons.
  • Added short Share Link URLs and Open on TV setup codes, so viewers can open a shared dashboard without typing a long URL, even with a TV remote. Existing password, email verification, disabled-link, and expiration rules still apply.

Improvements

• Dataset sync now runs each source connection as its own job, reducing interference between connections during refreshes.
• Raised the default Dataset processing capacity, so large analyses are less likely to stop early.
• Dashboards open and update faster, and calculation previews load more quickly.
• Metric Display Query Widgets now auto-size the primary value to fit the tile, with an optional fixed font size.
• Rolling date Filter Variable defaults now apply to the dashboard currently open.
• Removed the legacy Filter Variable configuration panel in favor of the current controls.
• Query Widget table footers now show Export beside Open Drilldown.
• Long Roar Designer runs now wrap up more cleanly when work can't be completed, and recover more reliably from stalled or interrupted sessions.

Bug Fixes

• Fixed nested calculation-column filters that could break when filter rows were reordered.
• Fixed false disconnect errors and offline-reload paths that didn't work on dashboards.
• Fixed previous-period category bars so matching labels pair correctly across periods.
• Fixed snapshot Dataset schedules and option saving that could leave schedules disabled or block option updates.
• Fixed Query Widget bar value caps so each bar rounds away from zero consistently.
• Fixed Query Widget filter refreshes interfering with table processing, and unnecessary repeated refreshes on Query Widgets that share a source.
• Fixed Query Widgets that dropped text cells or recent chart dates.
• Fixed scheduled and default Query Widget renders that failed to display after related query changes.
• Fixed template usage checks that could fail when templates were used at the same time.
• Fixed shared-dashboard drilldowns that could spin indefinitely after a failed request.
• Fixed Share Link email verification so recipients see a loading state and clear errors.
• Fixed Share Link filter state so each viewer's selections stay isolated.
• Fixed company access approvals so seat-limit and related refusals show a clear warning instead of failing.
• Fixed wide tables in Roar Designer chat so they scroll horizontally and stay selectable.
• Fixed Roar Designer date-filter handling for Dataset and widget work.
• Fixed company knowledge proposal approval controls so review actions work again.
• Fixed snapshot Dataset proposals with clearer dataset ID and ordering-column validation.
• Fixed in-dashboard Roar Designer messages so they continue in the selected chat instead of a different thread.
• Fixed Roar Designer SQL recovery so a recovered query is no longer blocked.
• Fixed knowledge and command searches that failed on common options or very large folder matches. Wide searches now return a limited, clearly marked result.
• ConnectWise: Fixed ticket audit trail previews that failed when audit timestamps included fractional seconds.
• Fixed product updates so a new build activates only after you accept the reload prompt.