Official client libraries for the Lingara API: generate vocabulary lists, create lesson plans, hold tutor conversations and read your usage, from your own code.
The libraries are pre-release: every version before 1.0.0 carries an
-alpha.N, -beta.N or -rc.N suffix, and every install line pins one.
| Language | Package | Status | Install |
|---|---|---|---|
| TypeScript | @lingara/api (npm) |
pre-release | npm install @lingara/api@next |
| Rust | lingara (crates.io) |
pre-release | cargo add lingara@<version> |
| Go | github.com/Spinning-Cat-Studios/lingara_api_clients/go |
in development | |
| Java | com.getlingara:lingara-java (Maven Central) |
in development | |
| Kotlin | com.getlingara:lingara-kotlin (Maven Central) |
in development | |
| Ruby | lingara (RubyGems) |
in development | |
| PHP | spinningcatstudios/lingara (Packagist) |
in development |
Each row gains its install line when its library is released. The released
libraries are the ones languages.toml lists; every release publishes them all
at one version, and each release's notes name the exact version to pin.
spec/openapi.json— the Lingara API's OpenAPI 3.2 document, withversions.tomland the frozen versions underspec/versions/.spec/SOURCEnames the commit it was taken from.spec/generator/— the generator view of that document, in OpenAPI 3.1 and 3.0. No mainstream generator reads 3.2, so the four event-stream operations are lifted into named, discriminated unions and listed underx-lingara-streams, and everything else is downconverted. The view is an input to code generators only; the API servesspec/openapi.json.tools/spec-codegen/— the Rust tool that writes the view. It refuses any construct it cannot express rather than approximating it.
make help # every target
make spec-view # regenerate spec/generator/
make check-publishable # the view is current (what CI runs)
cargo test --workspaceMIT — see LICENSE.