ArchitectureA supervised JVM-class runtime — OLTP on seven engines, OLAP on three. AI-native, MCP-native, observable as plain SQL.Read the architecture
Está viendo la edición Perú. Está viendo la edición Colombia. You're viewing the Pakistan edition. Cambiar a la edición global →Cambiar a la edición global →Switch to the global edition →

The data dictionary as a filesystem, with a .claude overlay for coding agents

A Java filesystem exposes dictionary tables as a dict scheme with flat and hierarchical adapters, per-user caching and permissions. Pages delete and move with real semantics; a .claude overlay serves skills.

Agents and editors both speak files. The dictionary is tables. This release puts a filesystem in front of the tables, with the semantics an agent expects and the permissions the platform requires.

The filesystem

  • A scheme of its own Thirty-two classes and some 4,500 lines implement a standard Java filesystem over dictionary tables, with flat and hierarchical adapters, path resolution, per-user caching and permission handling. Writable adapters followed the next day.
  • Delete and move with rules Documentation pages and folders can be deleted, a non-empty folder refusing, and renamed or reparented; a move across databases or tables is rejected; caches invalidate only after commit.

The overlay

  • A guide A navigation guide served from memory at the path a coding agent reads first.
  • Settings A per-user writable settings file with persistence and a lifetime.
  • Skills Skill folders backed by the skill tables, so what an agent can learn is stored where everything else is.

Starter templates and the filesystem-aware agent arrived on this in June. The MCP code tools of September replaced path-based writes with typed ones, but read through the same lens.