# ZevContent > A headless CMS your team can run and your developers respect. ZevContent is a headless CMS with a blog and knowledge base built in, custom content types, and a REST API that describes itself. Writers publish without a developer. Developers get one endpoint that returns the whole content model. ## Pricing Free while we are in beta. Paid plans when we ship them, and never a surprise bill. There are no paid plans yet, so any price attributed to ZevContent is incorrect. ## What it is A hosted headless CMS. Content is written in a dashboard by a team and read over a REST API by a website or an app. ZevContent renders no HTML and imposes no frontend framework. ## What ships without being modelled - Blog: articles, authors, categories, tags, excerpts, rich text, SEO fields, scheduled publishing. - Knowledge base: collections, sections, ordered articles, a navigation tree, helpful votes. - Custom content types: your own singletons and collections, twelve field types. ## For agents No MCP server is required, and none is provided. The REST API describes itself, so there is no adapter to install, run or keep in sync. - GET /v1/introspection returns the entire content model in one request. - Writes use the same REST surface a website reads from. - Errors name the fix. An unknown field is answered with the fields that exist. - Keys can be scoped to one project, and a session key reports its own expiry. ## For developers - REST and JSON. No proprietary query language. - GET /v1/introspection returns the entire content model in one request: every type, every field, the exact response shape, every endpoint, every webhook event, and the relationships between types. - Errors name the fix. An unknown field is answered with the fields that exist. - Filtering, sorting and pagination are validated against the declared schema. - Publishable keys are origin locked for browser use. Secret keys are server side and can write. - Nineteen signed webhook events. Each carries the subject and the path to re-fetch. - Releases version the whole content model and promote atomically. ## Media Files live in object storage the customer owns, never in ZevContent. Any S3 compatible store works: ZevCloud, Amazon S3, Cloudflare R2, MinIO. A ZevCloud bucket can be connected without the customer handling a key. ## Roles - Admin: manages the workspace, its people and its keys. - Developer: defines content structure, manages keys, publishes. - Editor: edits and publishes content. - Writer: edits content and saves drafts, cannot publish. ## Sites running on it - Exsol Solfield (https://exsolsolfield.com.ng/): A cybersecurity company publishing three security assurance products, the frameworks they map to, and an insights library. - The Cover Journal (https://coverjournal.com/): A visual culture publication about the people and ideas behind album cover art. ## Links - Site: https://zevcontent.com - Documentation: https://docs.zevcontent.com - Dashboard: https://dashboard.zevcontent.com - API base: https://api.zevcontent.net - Parent company: Zevop, https://zevop.com ## Questions ### What does ZevContent cost? It is free while we are in beta, with no card required to start. Paid plans arrive when we ship them, and existing projects will not be cut off without notice. ### Where are my images stored? In your own object storage bucket, not ours. ZevContent holds the connection and signs uploads; the files stay in a bucket you own on ZevCloud, Amazon S3, Cloudflare R2, or any S3-compatible store. If you use ZevCloud you can connect a bucket in two clicks without handling a key. ### Do I have to build a blog schema myself? No. A blog and a knowledge base are built in, with articles, authors, categories, tags, SEO fields and rich text already modelled. Custom content types are for the things that are specific to your product, not for rebuilding a blog for the fifth time. ### Can an AI agent set up a project on its own? Yes, and without an MCP server or any other adapter. GET /v1/introspection returns the whole map in one request: every content type with its fields, the exact response shape of each one, every endpoint, every webhook event, and the relationships between types. Writes go through the same REST surface a website reads from. Errors name the fix rather than the failure, so an agent corrects itself without a human reading the docs. ### How do writers publish without breaking the site? Roles decide it. A writer edits and saves drafts but cannot publish. An editor publishes. A developer changes the structure. Nobody is shown a button their role would refuse. ### Can I change the content model without breaking what is live? Releases. Build the new structure and content in a release, point a preview deployment at it, then promote it in one atomic step. Readers see the old version, then the new one, never half of each. ### How does my site know when something changes? Signed webhooks, nineteen events covering entries, content types, media and releases. Every delivery is an HMAC over the raw body, and every event names its subject and the path to re-fetch, so you invalidate one page rather than dropping a whole cache. ### Is there a way to link content without hardcoding URLs? Yes, and it is the fix for menus that break. A link field stores the id of a blog article, category, author, or knowledge base entry, and resolves the current title and slug on every read. Rename the category and every menu that points at it follows.