Most CMS connectors stop at writing content. Having `get_story_versions` and `restore_story` in the same tool set is what makes it reasonable to let an assistant edit a live space — a bad update is one call to undo. Releases are the other piece worth knowing about: scheduled batches mean a set of changes can be prepared and go live together rather than trickling out.
A Storyblok management server. It wraps the Management API so an assistant can create and publish content, organise it, and handle the release and webhook plumbing around it.
- Stories: `fetch_stories` with filtering, pagination and search, plus `get_story`, `create_story`, `update_story` and `delete_story`
- Publishing: `publish_story` takes it live and `unpublish_story` pulls it back
- History: `get_story_versions` lists previous versions and `restore_story` rolls one back; `import_story` brings a story in with an optional language code
- Tags: `fetch_tags`, `create_tag`, `delete_tag`, and `create_tag_and_add_to_story` in one step
- Webhooks: `fetch_webhooks`, `get_webhook`, `create_webhook`, `update_webhook`, `delete_webhook`
- Releases: `fetch_releases`, `create_release` and `update_release` for scheduled batches of content
The npm package storyblok-mcp, version 1.0.0. Clone it, build, and point your client at `build/index.js`. It needs STORYBLOK_SPACE_ID, STORYBLOK_MANAGEMENT_TOKEN and STORYBLOK_DEFAULT_PUBLIC_TOKEN, plus GOOGLE_GENERATIVE_AI_API_KEY for the AI-assisted parts.
Build from source — clone the repository and build it, then point your client at the binary