Gå til hovedinnhold

Reference

Generated, authoritative documentation for everything the framework exposes — commands, formats, tools, the project file, and the exchange format — derived directly from the code and the format and tool schemas. This page lists every reference and dashboard in one place; the same set is mirrored in the navbar's Reference menu.

Interactive references​

Generated from the live registries and schemas:

  • Kapi CLI commands — every kapi command, with runnable examples.
  • Format reference — every format reader/writer and its configurable parameters. Each card is tagged built-in (the native engine) or Okapi bridge (a filter provided by the Okapi bridge plugin).
  • Tool reference — every processing tool and its inputs and outputs, grouped by category and tagged by source the same way.

Each entry also has its own static, indexable page (see the note below).

Configuration & integration​

  • Project file — the .kapi recipe format, its fields, and how kapi discovers a project.
  • KLF format — the Kapi Localization Format, the JSON exchange serialization of the content model: spec, examples, an XLIFF comparison, and an interactive lab and conformance suite.
  • MCP server — the kapi MCP tools exposed to AI assistants.

Project status​

These dashboards are generated from the test and benchmark suites:

  • Parity — coverage against the Okapi Framework.
  • Format maturity — each format's position on the L0–L4 maturity rubric.
  • Benchmarks — performance measurements.
  • ML benchmark — measurements for the in-process ML models.
  • Check eval — evaluation of the verification checks.
  • Test results — the Okapi vs neokapi filter test comparison.
merknad

Every command, format, and tool also has its own static, indexable page under /reference/commands/<id>, /reference/formats/<id>, and /reference/tools/<id> — generated from the same data and linked from each card in the grids above. These per-entry pages are the canonical, shareable URLs; the grids remain the browse index.