Every change to devguard, dated and in plain language. We move your certified ISMS without breaking your evidence, and keep the platform under it steady, secure and improving. Here's the record.
Want updates like these on your own stack?
Start free, and every improvement lands in your workspace automatically: no upgrades, no migrations.
This release adds SAML single sign-on with directory sync as a Business add-on, security keys, and a two-factor requirement that is enforced for every credential.
Single sign-on and directory sync (add-on): sign in through your organization's SAML identity provider, require it for every member, and let SCIM create, update and deactivate members from your directory. The sign-in screen offers single sign-on before any code and sends an address under a required domain straight to its provider. See the separate announcement for the full picture.
Security keys: add a USB or NFC security key next to your passkeys under Account security.
Two-factor requirement: a sign-in through an identity provider, a passkey that checks a PIN or biometric, or an authenticator code meets your organization's requirement, which is now enforced on the server for sessions, API keys and connected apps alike. Admins see members' keys and grants without a second factor and can revoke them.
Fixed members: pending invitations are shown only to roles that may invite, and seats are counted once, in one place.
Fixed mandates: an organization whose own subscription is cancelled can be attached, clients are billed on the mandate's plan everywhere, attaching stops at a pending licence reduction, and a detach that would leave a client without members of its own is refused.
Fixed sign-in screens: the email address sits on its own line when you confirm a change, and the product updates checkbox lines up with its label.
Announcement
Your dashboard, arranged around your own work
Until now the dashboard was one layout for everyone: the same sections in the same order, whether you run the whole ISMS or sign off one audit a year. Each member now arranges their own. Open Customize dashboard beside Create, add the widgets you want from a library grouped by category, drag them into the order you read them in, and set the width of each so the important ones get the room.
The former tabbed sections are now single widgets, so your registers, work lists, evidence cards and alerts can each stand on their own. A new Next audit widget counts down to the audit under way or the next one planned and shows its open and overdue findings. Your layout is saved per organization, so your view of a client can differ from your view of your own company, and nothing you move changes a colleague's dashboard.
Nothing changes until you want it to. The familiar layout stays the default, and Reset to default brings it back at any time. Available on every plan.
This release adds access roles, a dashboard each member arranges for themselves, single sign-on settings, a templates gallery, policy packs and a calendar for deadlines, and it corrects how schedules handle the end of a month.
Access roles: owners and admins define roles with a level per area of the app (no access, view, edit or manage) and assign them to members. Lists, deadlines, search, notifications, the Trust Center and the assistant all follow them. Business plan. See the separate announcement for the full picture.
Customizable dashboard: each member arranges their own dashboard from a library of widgets, drags them into order and sets their width. A Next audit widget counts down to the audit under way or the next one planned. See the separate announcement.
Single sign-on settings: register your SAML identity provider from its metadata, require single sign-on for the organization, and see directory-managed members marked under People. Part of the single sign-on and directory sync add-on; see the separate announcement.
Templates gallery: browse templates as cards with a preview of each, from a process's structure and a policy's text to a questionnaire's questions. Search and filter them, or switch to a list.
Policy packs: the security templates become a pack of 16 ISO 27001 documents covering clauses 4 to 10 and the Annex A themes, with documents for ISO 9001, 14001, 45001 and 42001 alongside. Values several policies share become variables with a proposal in square brackets, an open field blocks approval, an approved version keeps the values it was approved with, and a policy can start from a template in the Add policy sheet.
Changelog
Released Device Agent Version 0.4.1
A maintenance release of the devguard agent.
Updater: the agent's update mechanism moves to a current archive library. What the agent checks and reports is unchanged.
This release brings DORA, NIS2 and the nFADP up to their current rules and adds checks for Microsoft Entra ID, Intune and devguard Devices.
Current rules: DORA carries the incident reporting clock of the 2025 regulations (4 hours after classification and at the latest 24 hours after awareness, then 72 hours and one month) and the subcontracting rules for critical or important functions. NIS2 carries the implementing regulation for digital infrastructure, managed service and digital providers with its thresholds for a significant incident. The nFADP notes that the FDPIC may pass a breach notification to the Federal Office for Cybersecurity with your consent. Identifiers are unchanged, so your coverage stays as it is.
CCM v4.0: the retirement date moves to the end of the CSA transition in December 2027.
devguard Devices: one check per device check. It passes only when every applicable device passes and opens a finding for each device that fails, cannot be read or is not reporting.
Microsoft Entra ID and Intune: both connect through admin consent to devguard's app, and one connection serves both. The Azure integration only ever called the same Microsoft Graph API and merges into Entra ID; its connections, checks and evidence move with it. The setup instructions name the Application permissions each check needs.
Announcement
Audits from plan to follow-up
Audits in devguard now cover the whole cycle, from planning the programme to closing the last finding. Pick an audit's scope down to individual controls and clauses, give it formatted objectives, criteria and a conclusion, and name the framework it audits. The audit programme counts from that scope and shows the progress of every clause, with markers where findings were raised.
Findings no longer start only in an audit. Raise one from a process run, an incident or a review item you reject, and it keeps the number of its source even after the source is gone. Nonconformities are a finding type of their own, and a finding can be linked to the control it concerns.
Follow-up stays with the people responsible. Assign a finding to a member or to a role, and the assignee is reminded before it is due, ahead of your admins. Tasks and comments on the finding record what was done, so the next audit sees the full story.
Scripts, pipelines and integrations can now use an organization token instead of a person's key. An organization token belongs to the organization itself: it runs as a service account that takes no seat, stays out of member lists and notifications, and keeps working when the colleague who set it up moves on. Admins create and revoke them in the organization settings.
Every token can be limited to the areas it needs. A scope picker sets read or write access per area, and personal tokens can now be bound to a single organization with their own scopes, so a key for one client or one job cannot reach anything else. The devguard CLI and the MCP server accept the new tokens, and the CLI keeps them out of its output.
Connected apps moved into the organization settings as well. Admins see every app connected to the organization, when it was last used, and revoke its access in one step.
This release adds organization tokens and tightens transport security across devguard.
Organization tokens: tokens for an organization run as service accounts that take no seat. Scoped tokens stay within their organization and the areas they were given, and admins list and revoke the apps connected to the organization.
Mandates: attach an existing organization you own or administer to a mandate.
Security: devguard answers on HTTPS only, with strict transport security and standard security headers.
Fixed accounts: you set up two-factor authentication on your own account before requiring it of the organization, the sign-in code arrives in your browser's language when no other is known, and a deactivated member can no longer be made owner.
Announcement
Run your quality management system in devguard
devguard now supports a full quality management system alongside your ISMS. Adopt ISO 9001:2026 and the platform brings what a QMS runs on: a quality policy, the processes ISO 9001 expects, records, supplier evaluation and a management review, all as built-in templates you adapt rather than write from scratch. The setup wizard recognizes the framework and preselects what fits.
Day-to-day quality work happens in processes. Templates cover customer complaints, nonconformity and corrective action, internal audits, supplier evaluation, contract review, design and development, release, change requests and checks of measuring equipment. KPIs carry targets and show their trend, the competence matrix shows which roles need which competences and who holds them, and opportunities sit in the register beside your risks. Calibration certificates and other documents with an expiry date appear in your deadlines before they run out.
When it is time for the management review, a report gathers the inputs of the period, from complaints and change requests to KPI results and audit findings, and links back to your review process. Controls, evidence and audits you already keep for information security count once and serve both systems.
This release takes devguard beyond information security to quality, AI, environmental and health and safety management, and adds opportunities, a competence matrix, incident reporting stages and organization tokens.
QMS: built-in policies, process templates, records, risk template sets and supplier questionnaires for quality, AI, environmental and health and safety management, including the ISO 9001 management review. A new Templates page lists those for your adopted frameworks first.
Setup wizard: it preselects what fits your frameworks, adds Processes and Records steps, asks about a planned external audit, and can be reopened from Frameworks and Settings.
Opportunities: record opportunities in the risk register, with their own responses, scale and matrix, and an opportunity report.
Management Review: a report gathers a period's inputs, including complaints and change requests, and is linked from your review processes.
Processes: set KPI targets and follow their trend, give reports a due date and a response time in hours, and run a checklist about a vendor or an asset from Submissions. A whistleblower template accepts anonymous submissions.
Competence matrix: define the competences each role needs and record completions as an admin.
Incident reporting: a Reporting tab lists each authority's reporting stages and their deadlines, which also appear in your calendar. Root cause and lessons learned take formatted text, and timeline entries can be marked internal.
Audits and findings: pick an audit's scope down to controls and clauses, and the audit programme shows progress per clause. Raise findings from process runs, incidents and rejected review items, assign them to a member or a role, and follow them up with reminders, tasks and comments. Nonconformities are a finding type of their own.
Policy import: when AI is on for your organization, it can read an imported PDF for you, and bold headings, shaded tables, page headers and wrapped dashes import better. A policy the assistant drafts arrives as sections, lists and tables, with its review schedule.
Internal audits: the scope picker shows the three-year certification cycle per clause, with what was audited, what is planned and what is not yet covered, a marker for recorded nonconformities and a filter for the gaps. Programme filters offer only existing programmes, and Close finding appears on a nonconformity once it is verified.
Deadlines: a calendar is now the default view, with a timeline and the list beside it. Both look 90 days ahead and pin overdue deadlines.
Schedules: build a review or check schedule from structured controls and preview its next five dates. A rule that can never occur is refused on save, and a monthly date past the end of a shorter month now falls on its last day. Stored review and due dates that had skipped a short month were corrected.
Microsoft Entra ID and Intune: connect both by signing in as a Microsoft administrator and granting consent, with one connection serving both. A check Microsoft refuses names the missing permission or licence.
Device checks: each device check now runs as an integration check that passes only when every device passes and opens a finding per failing or silent device. The daily Device fleet evidence records are retired and point to the integration.
Check runs: an expanded run shows the data it kept as proof, with copy and download.
Vendor questionnaires: pick a follow-up question's condition from the questionnaire's own questions and answers. Conditions survive copying a template, export and import.
Settings: shorter, reordered tabs. Usage sits on the Billing page, connected apps on Integrations, About in the footer, Create Organization beside your account's organizations, and the enterprise add-ons show their price. The owner can delete all data of an organization to start again, keeping the plan, the members and the audit log.
Mandates: members and auditors of a mandate reach only the clients a manager grants them on the new Access tab, People fits a mandate, and the template import preview says what a client's plan cannot hold.
Org chart: place a role on the chart and it follows whoever holds it.
Records: archive templates and categories from the UI and see what still links to them. Bulk delete takes only unfinalized records, and submitted checklists and resolved reports stay undeletable.
Notifications: emailed links open in the organization they belong to, a missed cycle of a recurring assignment reminds its member, a moved one-off deadline reminds again, and a notification opened from the dashboard is marked read.
AI: the organization's AI switch, now called AI features, also turns off questionnaire auto-fill, PDF import and embeddings. Only owners and admins write the knowledge base, and the limit banner shows your real usage and refreshes after each request.
Smaller improvements: IDs show in their own column the same way in every table, a finished review's title and notes can be edited, an attachment's full SHA-256 checksum can be copied, the risk overview lists overdue treatment actions once each, and every help page is rewritten as a manual with screenshots.
Fixed policies: the PDF keeps marks, nested lists, numbering and strikethrough, mentions resolve, approval requires a version in X.Y.Z form, and the approval dialog keeps its content until it has closed.
Fixed checks and actions: a check that reads an undeclared variable or an action with an unsupported placeholder is refused on save, and a paused check reactivates without re-judging its definition.
Fixed questionnaires: a library reuse counts once per item, the Owner filter nothing could set is gone, and reading a PDF questionnaire with AI counts against the quota.
Fixed processes: filed checklists leave the to-handle list, date answers are formatted in every submission view, runs are numbered by year on every surface, and a definition with duplicate ids or a choice field without options is refused.
Fixed Trust Center: download links, validity, revoke and preview states tell the truth, and the public page promises a request only for on-request certificates.
Fixed coverage: indirectly covered controls count in the legend, and the missing-evidence scan counts published policy sections and any link while skipping archived categories.
Fixed dates: CSV date-times without an offset are read in your timezone, the API accepts ISO 8601 only, and a day or month that rolls over is refused everywhere.
Fixed API: audit items are served at /audit-items/, a refused foreign id names the kind it expected, and a repeated link id is written once.
Fixed audit log: search finds items by their title and every item name, and entitlement reviews show as EAR-.
Fixed devices: an app match is checked before saving, with wildcards refused and plus signs accepted.
Fixed assets: removed class fields leave review snapshots, a deleted reviewed asset keeps its review item, and a rating of 9 or 10 reads Critical everywhere.
Fixed reports: a stalled logo no longer holds up a report, the SoA names the adopted framework, and a record report comes from the newest finalized version.
Fixed trainings: a source must be a link the training can open or embed.
Fixed elsewhere: done and cancelled tasks are never overdue, the roadmap axis labels month and year, losing access to an organization moves you to another instead of an error, audit logs survive a failed payment, German and English copy is corrected across the app, the account page reads well on phones and in both themes, and vendor websites are stored as https addresses.
Asset fields: each asset class carries its own typed fields, with presets and the next expiry date.
Suppliers: record shared responsibility, processing countries and the transfer mechanism, approve a supplier, and export the approved supplier list.
Trust Center: certificates can be downloaded on request, with manual approval and expiring links. Business associate agreements are a document type, and certificates show "Valid until".
Organization tokens: give a script or integration API access for the whole organization instead of a person. Personal tokens can be limited to one organization and to the areas you choose, and admins manage connected apps in the organization settings.
Integrations: connect Google Workspace and Google Cloud by signing in with Google. Stored credentials are encrypted per organization.
Automations: edit a check on a canvas, with drafts, version rollback and AI edits. Archive actions and checks, see failing checks in the Compliance Scan, and read an action's last error on its page.
Tables: every table has one toolbar with column filters, saved sorting and views, and new filters find overdue records and vendors and unanswered questionnaires. Dashboard alerts open the filtered list, tasks can be archived, and Mark as done sits beside Actions.
Policies: imports keep Word images and merged cells in PDF tables, and table cells can be merged and split. Move sections up and down, mention register entries by their ID, rename a policy, and link a control to a single section or the whole policy.
Controls and frameworks: record a rationale for each cross-link, carried through the framework CSV. Indirect coverage names what it rests on, CIS Implementation Groups mark the controls outside your group as not relevant in one step, and outdated framework editions sit behind a switch.
Smaller improvements: support opens from the user menu, form errors and dates appear in your language, and policies and Trust Center items can be reordered from the keyboard.
Fixed policies: saving without changes keeps a policy published, copied sections paste as sections, PDF exports keep merged cells, and imports join hyphenated words and drop repeated headers.
Fixed integrations: reconnecting GitLab, Asana or Jira keeps their webhooks and rotated tokens, Jira tickets go to the right assignee, Slack posts to the channel the action names, and GitHub checks use the right app installation.
Fixed vendors: a vendor with linked risks saves again, and a residual rating can no longer exceed the inherent one.
Fixed settings: a failed logo save shows the stored logo again, and ownership can only pass to an active admin.
Fixed devices: whether an operating system is current is judged from the vendors' release feeds on every platform.
Fixed CSV import: Excel exports are read correctly, an unreadable cell is named, and rows refused by the plan limit are explained.
Fixed lists and reports: lists refresh after every change, an active asset under an archived parent stays visible, the Statement of Applicability table is easier to read, and mentions render in PDF lists and tables.