The public website for ktext — a CLI tool that generates, validates, and exports CONTEXT.yaml, a machine-readable project context file for AI coding agents and developer tools.
Live at ktext.dev.
- Landing page — explains the problem ktext solves, how it works, and what the schema looks like in practice
- Getting Started — install instructions,
ktext init,ktext validate,ktext export, and CI integration - Schema Reference — complete field reference for all
CONTEXT.yamlsections - Scoring — how
ktext validatescores files across eight sections, section weights, quality checks, and JSON output - Common Questions — answers to "why not just use a README", "why not CLAUDE.md", and other common questions
- Astro 5 — static site, all pages prerendered
- Tailwind CSS 4 — utility-first styling with
@themedesign tokens - Cloudflare Workers — hosting via
@astrojs/cloudflareadapter
npm install
npm run dev # dev server at localhost:4321
npm run build # production build → dist/
npm run preview # preview the production build locallyThe site deploys automatically to Cloudflare Workers on push to main via .github/workflows/deploy.yml.
To deploy manually after a build:
npm run build
wrangler deploy --config dist/server/wrangler.jsonThe build generates the authoritative wrangler.json at dist/server/ — don't deploy from wrangler.jsonc directly.
Every codebase has context that lives in engineers' heads: why Postgres instead of MySQL, what you must never log, which layer owns what. When a new engineer joins, or an AI agent opens a PR, that context is missing. They guess. They get it wrong.
CONTEXT.yaml is a single file that captures that knowledge in a structured, machine-readable format. ktext generates it from your existing repo, scores it, and exports it to XML or JSON for efficient injection into any LLM context window.
The CLI is at github.com/arithmetike/ktext. MIT licensed, no accounts, no strings attached.
MIT