AI Schema Module Introduces Structured Drupal Data Exports for LLM Workflows

A new Drupal module, AI Schema, introduces a command-line approach to exporting Drupal entity models into structured formats designed for AI tools. Shared by Erdem Alkan in a LinkedIn post, the module is maintained by Erdem and Mete Berk Ergun, co-founders of OctoHaus, and provides a single Drush command to generate a complete representation of entities, fields, relationships and database mappings.

The module addresses a common challenge in Drupal development, where entity systems involve multiple field types, storage tables and relationships that are difficult to interpret outside the CMS. AI Schema converts this structure into simplified JSON outputs that can be consumed by large language models without requiring detailed knowledge of Drupal’s internal architecture.

Using the command drush aischema:export-all, developers can generate outputs that include entity and field definitions, relationship graphs between entities and SQL queries with correct JOIN logic. Outputs include files such as compact-model.json and relation-graph.json, alongside generated SQL queries, with per-entity files combining schema, relationships and queries into a single context.

The exported format is designed to reduce token usage while preserving semantic meaning. Fields are annotated with contextual hints such as labels, content or taxonomy references, allowing AI systems to interpret structure more effectively. This supports tasks such as SQL generation, migration planning and API scaffolding.

The module operates entirely through Drush and does not include a user interface. It requires Drupal 10 or 11 and Drush 12 or 13. The current release, 1.0.0-rc2, was published on 9 April 2026.

The announcement outlines the module’s capabilities through examples and command outputs but does not include implementation benchmarks or performance data.

The project is not covered by Drupal’s security advisory policy, according to its project page.

Further details and usage instructions are available on the project page.