Decap CMS vs Hygraph

GraphQL-native schema with union types and polymorphic relations. Content Federation adds external data sources.

Decap CMS supports the core field types you need: string, text, number, boolean, date, image, file, list, object, and relation. Nested structures are achievable via the object and list widget types. The ceiling appears on complex, deeply relational content models: the relation widget is limited to single-collection references, and there is no block-based component system comparable to Keystatic's blocks field or Sanity's portable text. For a blog or a marketing site with a defined content schema, it is sufficient. For a content platform with rich relational structure, the config will start working against you.

Component models embed across content types. Schema-first reuse is straightforward for GraphQL-experienced developers.

There is no native component or content block library. Reuse of content patterns across collections requires duplicating field definitions in config.yml, which becomes a maintenance burden as the schema grows. Partial YAML anchors can mitigate this, but it is a workaround rather than a feature. Compared to tools with explicit block registries (Sanity, Keystatic), the reuse story is weak.

Basic required, unique, and regex validators. Advanced custom validation is limited compared to alternatives.

Required fields, pattern matching via regex, min/max on lists, and basic type constraints are supported natively. There is no custom async validator system and no cross-field validation. For straightforward content models, the built-in validation covers most common use cases. Teams with business-rule-heavy validation requirements will need to handle that at the framework layer.

The editor is functional but complex. Editors need to understand GraphQL-style relationships before working efficiently.

The editorial interface is functional and not intimidating for non-technical users. A content editor can learn the basics within an hour: create an entry, fill in fields, upload an image, and save. The friction is on the conceptual model: saving creates a Git commit, and editors without any Git background occasionally find this confusing. The UI itself is clean but dated compared to Sanity or even Keystatic.

Preview via configurable URLs. No live preview panel, editors see changes after saving. Developer setup required.

Decap CMS includes a preview pane feature, but it requires custom React-based template configuration by a developer to render content as it would appear on the site. Out of the box, the preview pane shows raw field values rather than a rendered page. There is no visual in-context editing. For teams that need true live preview, the setup cost is non-trivial and the result is still not as polished as TinaCMS or Sanity Studio.

Custom content stages are configurable. Approval workflows with notifications are available on paid plans.

The editorial workflow feature (draft, in-review, ready states managed via Git branches) exists but is explicitly marked as beta and has a history of instability. In practice, when content changes involve more than a handful of files, merge conflicts can surface in ways that are hard for non-technical editors to resolve. For solo publishers or small teams with light workflow needs, it is usable. For any team that needs a reliable approval-before-publish chain, it is not dependable enough.

Handles uploads and basic transforms via the built-in API. Less capable than alternatives.

Media uploads are stored directly in the Git repository by default, which causes repo bloat on image-heavy sites. Cloudinary and Uploadcare integrations are available as media library options to offload asset hosting, but they require additional configuration. There is no native DAM, no image transformation pipeline, and no tagging or folder organisation at scale. Adequate for a small blog, limiting for a content-heavy site.

Presence indicators show concurrent editors. Live sync available but less polished.

No real-time collaboration. Simultaneous editing by two users on the same entry is likely to produce a Git conflict. There are no presence indicators, no inline comments, and no conflict-resolution UI. The collaboration model is the Git model, which works for developer teams and is an obstacle for dedicated content teams.

The most granular here, field, locale, content stage, and model-level access designed for complex enterprise orgs.

Access control is handled at the Git host level (GitHub, GitLab, Bitbucket) or via Netlify Identity. There are no collection-level or field-level permissions within the CMS itself. You cannot restrict an editor to a specific section of content or make certain fields read-only for certain roles. Adequate for teams where all editors have equivalent access; limiting for anything with role-based content ownership.

Multi-locale is first-class with field-level variants, locale-specific publishing, and multi-region API delivery.

There is no native multi-locale UI. The common workaround is separate collections per locale or a folder-based convention. An i18n configuration option exists in beta that enables locale-specific folders, but it is not a first-class feature and the documentation reflects its experimental status. Any project with serious localisation requirements should look elsewhere.

Native fallback configured in project settings with API-level enforcement. GraphQL returns fallback values automatically.

Locale fallback logic is not managed by the CMS. Teams relying on fallback behaviour must implement it entirely at the framework layer. There is no fallback indicator in the editorial UI and no mechanism to flag missing translations.

Excellent GraphQL docs with a live playground, schema introspection, and generated TypeScript types.

There is no delivery API. Content is read as files from the repository at build time. Decap CMS has no typed SDK, no GraphQL endpoint, and no REST API for content queries. Documentation is functional but reflects a project maintained primarily by volunteers: some sections are outdated, examples reference deprecated configurations, and TypeScript support is absent at the schema definition layer. A developer familiar with static site generators will find their way, but the experience is not polished.

Good framework examples. GraphQL-first means any client works easily, less framework tooling than alternatives.

Integration guides exist for Hugo, Jekyll, Gatsby, Astro, and Eleventy in the official docs. The Astro docs include a Decap CMS guide. Setup is manual: two files (index.html and config.yml) in a /public/admin directory, and you are running. No CLI scaffolding, no starter templates maintained by the Decap team. Third-party Astro starters include Decap CMS configurations. For Next.js, integration is possible but less documented than Hugo or Astro paths.

Full schema management via GraphQL mutations, types, relations, and fields all programmatically creatable.

There is no management API for programmatic content operations. Content is created and edited exclusively through the admin UI or directly as files in the repository. You cannot ingest content from external systems via API. Scripting is possible at the Git level, but this is not a supported workflow.

Branch schema and content for testing, then promote. Less polished than DatoCMS's one-click sandbox workflow.

Environments map to Git branches. You can configure Decap CMS to point at a different branch for a staging environment, but there is no first-class environment concept in the admin UI. Environment promotion is a manual Git operation. This is workable with developer discipline but requires establishing your own conventions.

Global CDN with multi-region data residency on Enterprise. API performance is strong globally, especially for GraphQL.

Content is read from the filesystem at build time, so there is no runtime API call and no CDN dependency for content delivery. This is the structural performance advantage of the Git-based model. With a fast static host (Netlify, Vercel, Cloudflare Pages), the full site including content is globally distributed at the CDN layer. External media libraries like Cloudinary add CDN-served image delivery.

Fully managed SaaS with zero infrastructure to configure or maintain.

There is no CMS server to deploy or maintain. The admin UI is a static HTML and JS bundle served from your /admin path. Netlify hosting is the path of least resistance (Git Gateway and Identity integrate directly), but Decap CMS works on any static host with any of its supported Git backends. No databases, no persistent servers, no CMS-side infrastructure bills.

Focused on e-commerce, frameworks, and enterprise tools.

The integration ecosystem is broader than Keystatic or TinaCMS for static site generators specifically. Hugo, Jekyll, Eleventy, Gatsby, Astro, and Hexo all have documented paths. Media library integrations with Cloudinary and Uploadcare exist. Backend support spans GitHub, GitLab, Bitbucket, Azure DevOps, and Gitea. The plugin and custom widget API allows extending the field type system. What the ecosystem lacks is momentum: fewer new integrations are being built compared to the Netlify CMS era.

Enterprise-niche community. Active Discord with responsive support but fewer tutorials and third-party resources.

With ~18,000 GitHub stars and a long track record, the project has significant visibility. The honest picture post-2023 is one of slower development: the rebrand from Netlify CMS to Decap CMS moved stewardship to a Slovenian agency (PM TechHub), and while the project is not abandoned, the release cadence has slowed compared to competitors. No pull request or issue activity was detected in recent months. The community remains active in discussions, but the direction-of-travel for product development is less clear than for Thinkmill-backed Keystatic or TinaCMS.