■ Features

The whole proof, page by page.

128 capabilities across 18 areas. 116 of them are in version 1.0; the rest are marked with the roadmap phase that brings them.

In 1.0Phase 3 On the roadmap

§ 01

Content modelling

Model content once as reusable pieces, and change the model whenever the business changes.

Content and element types: Document types for pages, element types for blocks, compositions to share field groups, singletons for one-off pages, and allowed-children rules for the tree.
Block List and Block Grid: Assemble pages from element types with separate content and settings, nested layouts in a grid, and presets editors start from.
A full set of field types: Rich text, pickers for content, media, members, and taxonomy terms, repeatable groups, links, embeds, locations, date ranges, tables, tags, and saved-query pickers.
Model changes at any size: Values are JSONB inside a relational shell, so changing a type on a 250,000-page site saves immediately: no server lock, no migration, no reindex.
Filterable fields, indexed: Fields marked filterable or sortable are materialized to indexed columns on save, so list queries stay fast without deep JSON scans.
Taxonomies and saved queries: Hierarchical vocabularies for tagging and navigation, and named queries editors can place in content and the Delivery API serves by alias.
Navigation menus: Menus built from the tree or by hand, per site and language, delivered as a single typed response.
Identity that travels: Every item has a global identifier that survives import and export, an optional unique alias for stable lookups, and optional short codes.
Blueprints and bulk editing: Start items from blueprints, and edit, move, tag, or find and replace across many items at once with a preview of every change.

§ 02

The editing experience

A backoffice designed like a press proof: calm, fast, keyboard-first, and accessible.

The sheet and the track: Each page is a sheet under a Draft → Review → Live track with one main button, a line of checks that opens a drawer, and comments in the margin.
TipTap rich text: Structured rich text stored as a document and delivered as both JSON and rendered HTML, with blocks, links, and media inside it.
Autosave and presence: Nothing is lost between saves, and editors see who else has an item open before they collide.
Comments, mentions, and tasks: Threads on content, media, translations, and approvals, with mentions that notify and tasks that can be assigned and closed.
Jump anywhere: The jump field (/ or Ctrl K) finds any page, place, or setting, and the Index (G) lays out every room of the backoffice.
Six rooms: Today, Content, Library, Audience, Build, and Settings group the work by what people do, with Today as the dashboard's face.
Tree, list, and board views: Browse a tree built for 250,000 items by loading each level on demand, filter it as a list, or move work across a board.
Shareable previews and PDF: Password-protected preview links that expire, preview as a segment or member, and a PDF export of any item for sign-off.
Accessible and multilingual: WCAG 2.2 AA with keyboard-only operation, right-to-left layouts, a night theme, and dates shown in each user's own timezone.

§ 03

Publishing governance

Every change is attributable, reviewable, schedulable, and reversible.

Approval workflows: Multi-stage approvals per content type or section, with users and groups at each stage, and rejection with reasons.
Scheduled publishing: Publish and unpublish items or whole sections at a time in any timezone, correct across daylight-saving changes, and fired exactly once.
Releases: Bundle many items into a release, preview the site as it will be, approve it, publish it atomically, and roll it back as one.
Versions and rollback: Immutable versions of every item, with rollback to any earlier version and retention policies that decide how many to keep.
Audit history: Who changed what and when, for content, media, settings, keys, and plugins, recorded on every mutation.
Locks with expiry: Editors can lock items they are working on; locks expire, and administrators can release them with a record of who did.
Recycle Bin with retention: Deleted items and media wait in a Recycle Bin, restorable with their place in the tree, until a retention policy empties it.
Automation rules: When something happens, if conditions hold, then act: triggers from content, members, schedules, and cron, with a test mode, run log, and versions.
Review dates and stale content: Give content a review date and a report of what is due or stale, so nothing out of date stays live unnoticed.
Notifications everywhere: In-app, email, and digests, plus Slack, Teams, and webhook channels, with subscriptions per node and templates you can override.

§ 04

Media

Cloudinary-style media handling built into the platform, served from your CDN.

URL-based transformations: Width, height, crop, focal point, format, and quality set from the URL, with automatic AVIF or WebP and signed URLs for anything unusual.
Named presets and responsive sets: Define renditions per site once, and deliver srcset-ready responsive images with the focal point and crops editors chose.
Video and audio: Transcoding to adaptive streams, poster frames, caption tracks per language, and renditions generated in the background.
Direct-to-storage uploads: Large files go straight to storage in resumable parts, with metadata extraction and malware scanning before anything is published.
Storage you choose: A local volume, S3-compatible, Azure Blob, or Google Cloud Storage per instance, with site and folder overrides and migration without changing a URL.
In-browser image editing: Straighten, rotate, flip, crop, and set the focal point in the backoffice, saved as a new version of the asset.
Rights management: Licence, credit line, releases, allowed uses, regions, sites, and expiry on every asset, checked when a page using it is published.
Bulk import and bucket sync: Import zip archives with a report, or keep a bucket or mounted folder in sync with the library by hand or on a schedule.
Usage and bandwidth: See where every asset is used before you replace or delete it, and bandwidth and cache hit rates per site.
AI-assisted mediaPhase 3: Automatic tags, alt text, subject detection for crops, background removal, and transcription.

§ 05

Delivery and performance

A versioned, cache-first Delivery API that stays fast on the busiest day.

The Delivery API: Content, media, search, menus, taxonomies, dictionary, redirects, sitemaps, feeds, forms, and widget configuration, in one consistent envelope.
Three cache layers: Edge cache keyed on the publish version, a tag-evicted output cache on every node, and Redis for coordination: each with one job and one way to invalidate.
Publish in under a second: Measured at 555 ms at the 95th percentile from publish to visible at the origin, against a 5-second target.
Degrades safely: If Redis goes away, nodes serve from local caches; if search goes away, PostgreSQL answers; if the origin goes away, the CDN serves stale.
Preview and staging hostnames: Preview tokens scoped to a subtree, draft mode for your frontend, and staging hostnames that show approved drafts.
Keys with scopes and limits: Public site keys bound to domain lists for the browser, and server keys with capability scopes and tiered rate limits for builds.
Signed webhooks: Per-site webhooks signed with rotating secrets, retried from a durable queue, logged, and redeliverable.
Usage reporting: Requests per key, hour, and outcome, plus the paths that returned not found, so you can fix links before visitors notice.

§ 06

Sites, domains, and URLs

Many sites and languages from one tree, with URLs that never break.

Hostnames per start node: Any number of hostnames per site with language prefixes, primary and alias roles, and a platform hostname for every site.
Visibility by hostname: Show or hide a branch on specific hostnames, and limit which sites each editor can work on in the backoffice.
Geofencing: Restrict content by country from the CDN's country header. No IP database, and location is never resolved beyond country.
Redirects and URL patterns: Moves and renames record redirects automatically; manage the rest by hand, with patterns that shape URLs per content type.
Short links and QR codes: Short codes that follow an item wherever it moves, with QR codes, expiry, and click counts by day, referrer, and country.
Sitemaps and robots: Sharded page, image, and video sitemaps maintained on publish and tested to 200,000 entries, with robots.txt per site.
Feeds: RSS, Atom, and JSON Feed for any section or saved query, with the right alternate links in every page's head.

§ 07

Translation and localization

Every language a first-class citizen, including right-to-left and CJK scripts.

Translation families: Translations are children of their master item, so structure and shared blocks stay in step while each language has its own state.
Status and outdated flags: When the master changes, its translations are flagged outdated field by field, and a translation can publish with its master.
Proofing and review: Translations go through their own review units, with comments in the margin and approval before they go live.
XLIFF round trip: Export to your translation agency as XLIFF and import the result, with memory from earlier translations applied.
Language fallback: Up to 50 languages per site with fallback chains, per-language hostnames or prefixes, and a translated dictionary.
AI translationPhase 3: Machine translation with a glossary and translation memory that editors proof before publishing.

§ 08

SEO, search, and site health

Findable pages, fast search, and a site that tells you when something breaks.

The SEO tab: Title, description, canonical, robots, Open Graph, and social cards on every item, filled automatically from the content by rules you set.
Structured data: Schema.org templates per content type, delivered as JSON-LD with breadcrumbs and language alternates.
Checks that can block a publish: SEO and accessibility checks on every item, shown on the track, with the option to stop a page going live until they pass.
Site monitoring: A scheduled crawl for broken links and missing metadata, uptime checks, and Core Web Vitals from real visitors with alerts.
Search you can tune: PostgreSQL full-text search out of the box or OpenSearch and Elasticsearch at scale, with synonyms, stopwords, weights, pins, and a report.
AI recommendationsPhase 3: Suggested titles and descriptions, and semantic, hybrid, and image similarity search.

§ 09

Personalization

Tailored pages without giving up the cache, and never without a default.

Segments: Audiences built from roles, profile fields, interests, country, language, device, referrer, campaign, time, and consent, combined with all, any, and not.
Block and field variants: Alternative content per segment on a block or a short field, kept in one item instead of duplicated pages.
Recommendation slots: Content-based, interest-based, trending, collaborative, and editorially pinned recommendations, blended by weight, with a fallback.
Promotions and restricted content: Promotion slots and items limited to a segment, alongside public access, hostname visibility, and geofencing.
Cache-safe by design: Decisions come from a separate uncached endpoint and are applied client-side by the loader, so content responses are identical for every visitor.
Anonymous first, consent-gated: Anonymous interest profiles live in the browser, behaviour-based personalization needs consent, and signed-in profiles merge on sign-in.
Preview as and reporting: Preview any page as a segment, a member, a country, or a future date, and report on which variants were shown.
ExperimentsPhase 3: A/B and multivariate tests, auto-optimization, AI-generated variants, and segment suggestions.

§ 10

Analytics

A first-party analytics store, the providers you already use, and your data warehouse.

First-party collector: Page views, clicks, downloads, scroll depth, and searches collected by the loader, consent-aware, into a store you run.
Provider plugins: Matomo, Google Analytics 4, and Microsoft Clarity bundled as plugins, configured per site without touching the frontend.
Content and editorial analytics: Performance per item and content type next to the content itself, and reports on the editorial process: throughput, time to publish, and approval turnaround.
Alerts: Thresholds on traffic and Core Web Vitals that notify the right people through the channel they chose.
Data warehouse export: Daily Parquet exports of analytics and, when you allow it, form data, for your warehouse or lake.

§ 11

Frontend users and forms

Accounts, sign-in, and forms for your visitors, without another identity product.

User pools: Pools per site, instance, or organization, so a visitor registers once and signs in everywhere in the pool with single sign-on across domains.
Registration and sign-in: Hosted registration, email verification, password reset, magic links, invitations, social and enterprise identity providers, and approval queues.
Profiles and self-service: Profile fields you define, an account widget for visitors, sign-in history, and explicit preferences that feed personalization.
Roles, groups, and public access: Frontend roles and groups with expiry dates, and member-only content that the Delivery API never leaks to anonymous callers.
Export and erasure: Data export and erasure on request, with a 30-day undo window and coverage checked for every table that holds a member's data.
Form builder: Versioned forms with sections, steps, conditional logic, validation, and file uploads, and a submissions view with assignment and states.
Encrypted submissions: Answers are encrypted with AES-256-GCM under a per-instance key and deleted on a retention schedule.
Spam protection and actions: Honeypots, timing, rate limits, and optional proof of work, plus notifications and webhook actions per form.

§ 13

Widgets

Functionality you maintain once and inject into every site with one script tag.

One loader: A single script tag with the site's public key registers the site's widgets as custom elements and loads each lazily.
Pinned, immutable versions: Each widget version is immutable on the CDN, pinned per site, and upgraded or rolled back with an audit record.
Built-in widgets: Consent banner, cookie declaration, account, and personalization come with the platform and share the loader.
Widget BuilderPhase 4: Author widgets in the backoffice with a code editor, live preview, server-side builds, tests, and a low-code mode.
Component CatalogPhase 4: A Storybook-style catalog inside the app that tracks component versions and variations per site.
Visual editingPhase 4: Edit on the live site through content source attributes and a backoffice overlay.

§ 14

Developer experience

Typed from the API down, scriptable from the pipeline up.

Consumer SDK: @modularcms/delivery, generated from the same OpenAPI document as the API, for any JavaScript runtime.
Next.js adapter: @modularcms/next: draft mode, revalidation, images, rich text, blocks, SEO metadata, menus, sitemaps, feeds, member sign-in, and personalization.
The modularcms CLI: Structure, packages, staging promotion, snapshots, migrations, keys, webhooks, jobs, and type generation, as a Node package, a single binary, or an image.
Structure as code: Keep content types and settings in files, plan and apply changes, and detect drift between the files and an instance.
API explorer: The Management and Delivery references inside the backoffice, with Try it and curl, fetch, and SDK snippets using your own keys.
Reference site and demo content: A Next.js consumer site and installable demo sites to start from and test against.

§ 15

Integrations and extensibility

Plug the platform into what you already run, and extend it without forking.

Git sync: Mirror content and structure to Git, two-way for structure, with plain Git or GitHub, release branches, and pull requests merged on publish.
External data: Reference records from commerce and other systems with an External Reference field that shows their current state in the picker.
Scheduled ingestion: Pull records from feeds and connectors into content on a schedule, with cursors, reports, and a named actor in the audit trail.
Marketing platforms: Brevo and Mailchimp plugins turn content into campaign drafts and sync member contacts and engagement.
Plugin packages: Install signed plugins per instance from publishers on a platform allow list, loaded in-process with rolling restarts.
Backoffice extensions: Plugins add screens, field types, and endpoints to the backoffice as custom elements, loaded with subresource integrity.
The ModularCMS StorePhase 5: Publish, discover, license, and sell packages in a public marketplace.

§ 16

Operations and data

The tools to run many sites calmly: back up, stage, promote, restore.

Snapshots and restore: Encrypted snapshots of an instance with its media, restorable to the same instance, a staging instance, or a new one.
Two kinds of staging: A separate instance for structural changes you promote with a plan, and staging hostnames for approved drafts.
Packages: Export and import content or structure as packages with a plan before apply and rollback after.
Umbraco migration: Import an Umbraco 13 to 17 site, Umbraco Cloud included, with a reviewable mapping, a rehearsal, and a report.
Maintenance mode and publish freeze: Pause editing or publishing per instance during upgrades, migrations, and restores.
Background jobs you can see: Every long task is a job with progress, cancellation, and fairness across instances, visible in an operations strip.
Upgrades: Semantic versions, rolling zero-downtime upgrades, a snapshot before every upgrade, and an opt-in update check.

§ 17

Security and tenancy

Isolation enforced by the database, and a supply chain you can audit.

Row-level security: Every tenant table leads with the instance identifier and is enforced by PostgreSQL row-level security, tested on every repository and endpoint.
Keycloak sign-in: OpenID Connect for backoffice users with MFA, and roles, grants, and per-node permissions held by the CMS.
Granular access: Grants limited to sites and languages, node permissions, and Management API keys whose rights never exceed their owner's.
Hardened by default: A strict hash-based Content Security Policy, security headers everywhere, rate limits on expensive operations, and controlled egress.
Verified supply chain: SBOMs for every image, Trivy scans with zero critical or high findings, and ZAP scans of the backoffice and both APIs.
Observability: OpenTelemetry traces, metrics, and structured logs with instance and request identifiers, and no personal data.

§ 18

AI content workflows

Coming in Phase 3: AI with your own keys, a budget, and an audit trail.

Bring your own keysPhase 3: Organization keys or platform-managed keys with metered billing, budgets, and cost reporting, through a provider abstraction.
Prompt managementPhase 3: Versioned prompts and prompt sets, shared across instances and packaged for reuse.
Editor assistancePhase 3: Drafting, rewriting, summarizing, and SEO suggestions inside the editor, with every call metered.
Agentic workflowsPhase 3: Multi-step content processes that draft, check, translate, and route to review.

■ Ready for press

See your content set in modules.

We’ll walk through the backoffice with your own site in mind, or import a copy of your Umbraco site and show you the result.

Book a demo Read the developer tour