CKEditor AI On-Premises overview

CKEditor AI On-Premises is the CKEditor AI backend packaged as a Docker image, so you can run it on your own infrastructure or in a private cloud. These articles cover what to prepare, how to install and configure it, and how to operate it.

If you want to know what CKEditor AI does and which integration fits your product, read the CKEditor AI guides first.

Read Architecture first, then the Install articles in order. The others stand alone:

• Plan – Architecture for how the parts connect and what to replicate when you scale.
• Install – Requirements, the Setup wizard, the Quickstart checklist, Deployment, and SSL.
• Configure – Required configuration for secrets, databases, and storage, then one article per area: LLM providers, MCP tools, Hooks, Web search, Web resources, Content moderation, and Guardrails.
• Operate – Observability, Logs, and Nightly releases.
• Reference – Changelog for what changed in each release.

CKEditor AI On-Premises runs under any Open Container runtime, such as Docker, Kubernetes, Amazon Elastic Container Service, or Azure Container Instances. A deployment is the application containers and the data stores they need. Requirements lists the infrastructure to prepare: a container runtime, an SQL database, an in-memory data store, and file storage. You also provide:

• Token endpoint – an endpoint in your application. It authenticates the user and returns a signed JWT, and every request to CKEditor AI On-Premises sends that token. Authentication explains the token. Create the token endpoint in Deployment shows where the endpoint fits in the installation.
• TLS termination – a load balancer or reverse proxy in front of the deployment, with your certificate installed on it. The application port serves plain HTTP. See SSL communication.
• LLM provider – an account with a cloud provider, or models you host yourself. You need at least one. To configure it, see LLM providers.
• MCP servers (optional) – your own MCP servers. CKEditor AI calls their tools during a request, and administrators can attach the resources of an MCP server to a context as files. See MCP tools.
• Collaboration Server On-Premises (optional) – if you run it, CKEditor AI On-Premises can share its SQL database, in-memory data store, environments, and management panel. See Collaboration Server integration.

• Requirements – Check whether your infrastructure can run CKEditor AI On-Premises.
• Setup wizard – Generate a docker run command for your deployment.
• Quickstart checklist – Track what to prepare, then send a first request to the deployment.
• Deployment – Install and run CKEditor AI On-Premises.