Skip to main content

Kapi CLI recipes

Each recipe takes a single task — estimate a job, catch UI truncation, enforce a glossary — states the goal, walks the steps, and lets you run the commands yourself. The interactive embeds run the real kapi binary in your browser: no install, no server, no API key. Every command shown is the command that runs.

For what kapi is and how the CLI and Kapi Desktop relate, see the Kapi overview. For every command and flag, see the command reference.

Ad-hoc or project?​

A .kapi project is the recommended working model for any repository or content set you localize repeatedly. It captures the languages, content patterns, and flows in a committed recipe once, adds git-style upward discovery of that recipe from any subdirectory, and keeps a project-local translation memory and termbase that every run reuses — so you stop repeating flags and start accumulating work. Create your first project walks the setup; the project file reference covers the recipe format and discovery rules.

kapi also runs ad-hoc: a command operates directly on the files you name, configured by flags, with nothing saved — kapi ai-translate -i messages.json --target-lang fr. That's the quick path for one-off jobs, scripting, and CI, and a fine on-ramp. Many recipes here are shown ad-hoc because they're genuine one-offs; each notes the project-mode equivalent where it helps.

Every recipe is marked at the top with its mode — Runs anywhere (ad-hoc) or Needs a .kapi project.

Recipes​

Get oriented in kapi

The top-level commands, format readers, and processing tools — a 30-second tour to find your way around.

Estimate a translation job

Count blocks and source words in any supported format to size up a quote before you commission it.

Catch UI truncation early

Pseudo-translate your strings to surface clipping and missing-string bugs before real text arrives.

Where TM and termbases live

The three storage models — loose file, named store, or project store — and the flags that select each.

Enforce terminology

Import a glossary and flag terminology drift in translated files before they ship.

Prepare content for translation

Protect sensitive content, reuse past translations, and keep terminology and quality consistent — one repeatable pass before anything ships.

Pre-translate with TM + termbase

Leverage existing TM, fill the gaps, and pre-flag terminology — deterministic, no API key, before any vendor sees content.

Run a bilingual round-trip

Extract to XLIFF, hand it to a translator, merge the targets back — with full TM provenance. Project-based.

Work in a .klz workspace

A single-file, serverless workspace: extract sources in, run tools in place, merge the finished files out — config travels with the file.

Modes & bindings

How ad-hoc and project runs differ, the source/sink binding model, and when it makes sense to mix them.

Next steps​