Use developer, IDE, and AI authoring tools

Changelogs are plain text files. There is no required authoring tool. Your team writes them in the same editors and IDEs it uses for application code, and they move through the same version control workflow.

Editors and IDEs

Any editor works. For modeled changelogs, XML files reference the Liquibase schema, so an IDE that understands XML schemas can validate structure and suggest change types and attributes as you type. YAML and JSON changelogs get the same benefit from your editor's general linting. For SQL changelogs, your existing SQL tooling applies unchanged.

Whatever the editor, the check before commit is the same. Run validate to catch malformed changelogs before they reach the pipeline. See Validate and preview a change.

Generate rather than write

Two built-in generators produce changelogs for you. The init command starts a project with a working changelog, and generate-changelog builds one from an existing database. See Create your changelog from an existing database and Let generate-changelog build your structure.

AI-assisted authoring

The Liquibase AI Changelog Generator drafts changelogs with AI assistance. It is optional tooling a team can adopt, and nothing in the workflow requires it. See the AI Changelog Generator documentation.