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 →

A static review of a dictionary's SQL: classify, prove, explain and measure

Quality tools classify every statement, fingerprint duplicates, prove each by prepare only, apply twelve rules served as data, scan the schema for drift and missing indexes, and read plans unrun.

A dictionary with a thousand reports has a thousand statements nobody has read in years. This release gives an agent, or an engineer, a set of quality tools that read them all without executing one.

Statements

  • Classified and measured Every report and view statement is classified by kind, measured for joins, subselects, unions, nesting, SELECT star and ORDER BY, and fingerprinted on normalised SQL so duplicates group together.
  • Proven, not run Each statement is proved by prepare only. Nothing executes.
  • Twelve rules as data Each with an identifier, a severity, a why and a fix, one detector each, served as a resource with a review guide beside it.

Schema and plans

  • Schema scan Engine metadata is read for type drift, name variants, foreign keys without an index, which Informix creates and PostgreSQL does not, foreign-key type mismatches and missing primary keys.
  • Explain without executing The plan is obtained without running the query, typed dummy values substituted for binds, and read with a rule set per engine for Informix, PostgreSQL, MySQL, Oracle and SQL Server.

Columns

  • Every CLOB measured in characters and bytes against the driver's ceilings: of 76 large-object columns in the platform's own dictionary, 16 were justified, 28 fit an in-row string and 32 were empty. About 700 values a minute over a remote link.
  • Declared width against stored A width scan compares each column's declared width to its longest value.

The tools are read-only by class and answer in the same shape whoever asks, so the review an agent runs on Monday is the review an auditor can rerun on Friday.