About SparkLang

SparkLang is the public name for the Spark programming language at sparklang.dev. A language beats another Python SDK because one reviewable .spark file covers train → eval → ship, with dry-run fixtures for CI without API keys.

What it is

A language + runtime for plain .spark files.
Dry-run / offline by default — no API keys and no network for learning and CI.
Real training via model train / model build (jobs + status). model plan is plan-only markdown.
Optional live ask against any OpenAI-compatible gateway via AI_GATEWAY_URL and ./spark --live.

Name note

Unrelated to Apache Spark (big data) and AdaCore SPARK (formal methods).

Core vs optional

Surface	Role
Core	Language, dry-run fixtures, model train/build/status, ask/eval helpers, playbooks, IDE language ops
Optional	Live ask/embed; voice (gated); browser / capture

Trust

License: MIT
Version: 0.6.3 — Changelog · Downloads
Source: github.com/sparklang-dev/sparklang (public MIT). Installers ship from sparklang.dev.
Contact: hello@sparklang.dev · noreply@sparklang.dev