AI agents, workflow builders, low-code tools and decoupled frontends all ask Drupal the same question: what can you do, and how do I call it? Today every integration answers it differently. MCP servers, FlowDrop, ECA, Maestro and custom modules each describe Drupal's capabilities in their own way.
The Rosetta Sprint, announced in the DriesNote at DrupalCon Rotterdam 2026, aims to fix this at the core: Drupal describes each capability once, consistently and machine-readably, through the Tool API, so agents, workflow systems, frontends and people can all use the same definitions.
In this lightning talk:
- The problem: why "one capability, many descriptions" holds Drupal back in the AI era
- The idea: typed, JSON Schema-based tool definitions, kept separate from how they are invoked
- Where the Tool API is now on its way to a stable release, and the design decisions being made
- Lessons from FlowDrop, which has used schema-typed inputs and outputs in production
- How to get involved: issues, reviews, and modules that need tool definitions
This is a strategic topic for the whole community. The more modules describe their capabilities, the more useful Drupal becomes for every AI and automation tool built on top of it.