kern-lang 1.0.0 → 2.0.0

package/dist/index.d.ts

@@ -1,28 +1,17 @@
 /**
- * Kern — the LLM-native language.
+ * kern-lang v2.0.0 — compatibility wrapper
  *
- * Swiss-engineered IR that transpiles to React, Next.js, Express,
- * Tailwind, and React Native.
+ * Re-exports all @kernlang/* packages for backwards compatibility.
+ * For new projects, import from @kernlang/core + @kernlang/<target> directly.
  */
-export { parse } from './parser.js';
-export { decompile } from './decompiler.js';
-export type { IRNode, IRSourceLocation, SourceMapEntry, TranspileResult, DecompileResult, GeneratedArtifact, KernEngine, } from './types.js';
-export { resolveConfig, mergeConfig, DEFAULT_CONFIG, VALID_TARGETS } from './config.js';
-export type { KernConfig, KernTarget, ResolvedKernConfig } from './config.js';
-export { transpile } from './transpiler.js';
-export { transpileWeb } from './transpiler-web.js';
-export { transpileTailwind } from './transpiler-tailwind.js';
-export { transpileNextjs } from './transpiler-nextjs.js';
-export { transpileExpress } from './transpiler-express.js';
-export { transpileCliApp } from './transpiler-cli.js';
-export { transpileTerminal } from './transpiler-terminal.js';
-export { collectLanguageMetrics, mergeMetrics, isEscapedStyleKey } from './metrics.js';
-export type { LanguageMetrics, StyleMetrics, NodeTypeMetrics } from './metrics.js';
-export { scanKernProject, projectToKern } from './context-export.js';
-export type { ProjectSummary } from './context-export.js';
-export { buildKernDraftPrompt, parseKernDraft, buildKernRankPrompt } from './draft-protocol.js';
-export type { KernDraft } from './draft-protocol.js';
-export { KERN_VERSION, NODE_TYPES, STYLE_SHORTHANDS, VALUE_SHORTHANDS } from './spec.js';
-export { stylesToTailwind, colorToTw, pxToTw, DEFAULT_COLORS } from './styles-tailwind.js';
-export { expandStyles, expandStyleKey, expandStyleValue } from './styles-react.js';
-export { countTokens, serializeIR, camelKey, escapeJsx } from './utils.js';
+export * from '@kernlang/core';
+export * from '@kernlang/protocol';
+export { transpileTailwind, transpileNextjs, transpileWeb } from '@kernlang/react';
+export { transpile } from '@kernlang/native';
+export { transpileExpress } from '@kernlang/express';
+export { transpileCliApp } from '@kernlang/cli';
+export { transpileTerminal } from '@kernlang/terminal';
+export { collectLanguageMetrics, mergeMetrics, isEscapedStyleKey } from '@kernlang/metrics';
+export type { LanguageMetrics, StyleMetrics, NodeTypeMetrics } from '@kernlang/metrics';
+export { scanKernProject, projectToKern } from '@kernlang/metrics';
+export type { ProjectSummary } from '@kernlang/metrics';

package/dist/index.js

@@ -1,33 +1,20 @@
 /**
- * Kern — the LLM-native language.
+ * kern-lang v2.0.0 — compatibility wrapper
  *
- * Swiss-engineered IR that transpiles to React, Next.js, Express,
- * Tailwind, and React Native.
+ * Re-exports all @kernlang/* packages for backwards compatibility.
+ * For new projects, import from @kernlang/core + @kernlang/<target> directly.
  */
-// Core
-export { parse } from './parser.js';
-export { decompile } from './decompiler.js';
-// Config
-export { resolveConfig, mergeConfig, DEFAULT_CONFIG, VALID_TARGETS } from './config.js';
+// Core (parser, types, config, spec, styles, codegen, utils)
+export * from '@kernlang/core';
+// Protocol (draft protocol for AI engines)
+export * from '@kernlang/protocol';
 // Transpilers
-export { transpile } from './transpiler.js';
-export { transpileWeb } from './transpiler-web.js';
-export { transpileTailwind } from './transpiler-tailwind.js';
-export { transpileNextjs } from './transpiler-nextjs.js';
-export { transpileExpress } from './transpiler-express.js';
-export { transpileCliApp } from './transpiler-cli.js';
-export { transpileTerminal } from './transpiler-terminal.js';
-// Metrics
-export { collectLanguageMetrics, mergeMetrics, isEscapedStyleKey } from './metrics.js';
-// Context export (for Agon integration)
-export { scanKernProject, projectToKern } from './context-export.js';
-// Draft protocol (for Agon forge/brainstorm/tribunal)
-export { buildKernDraftPrompt, parseKernDraft, buildKernRankPrompt } from './draft-protocol.js';
-// Spec
-export { KERN_VERSION, NODE_TYPES, STYLE_SHORTHANDS, VALUE_SHORTHANDS } from './spec.js';
-// Shared style engines
-export { stylesToTailwind, colorToTw, pxToTw, DEFAULT_COLORS } from './styles-tailwind.js';
-export { expandStyles, expandStyleKey, expandStyleValue } from './styles-react.js';
-// Utilities
-export { countTokens, serializeIR, camelKey, escapeJsx } from './utils.js';
+export { transpileTailwind, transpileNextjs, transpileWeb } from '@kernlang/react';
+export { transpile } from '@kernlang/native';
+export { transpileExpress } from '@kernlang/express';
+export { transpileCliApp } from '@kernlang/cli';
+export { transpileTerminal } from '@kernlang/terminal';
+// Metrics + context export
+export { collectLanguageMetrics, mergeMetrics, isEscapedStyleKey } from '@kernlang/metrics';
+export { scanKernProject, projectToKern } from '@kernlang/metrics';
 //# sourceMappingURL=index.js.map

package/dist/index.js.map

@@ -1 +1 @@
-{"version":3,"file":"index.js","sourceRoot":"","sources":["../src/index.ts"],"names":[],"mappings":"AAAA;;;;;GAKG;AAEH,OAAO;AACP,OAAO,EAAE,KAAK,EAAE,MAAM,aAAa,CAAC;AACpC,OAAO,EAAE,SAAS,EAAE,MAAM,iBAAiB,CAAC;AAa5C,SAAS;AACT,OAAO,EAAE,aAAa,EAAE,WAAW,EAAE,cAAc,EAAE,aAAa,EAAE,MAAM,aAAa,CAAC;AAGxF,cAAc;AACd,OAAO,EAAE,SAAS,EAAE,MAAM,iBAAiB,CAAC;AAC5C,OAAO,EAAE,YAAY,EAAE,MAAM,qBAAqB,CAAC;AACnD,OAAO,EAAE,iBAAiB,EAAE,MAAM,0BAA0B,CAAC;AAC7D,OAAO,EAAE,eAAe,EAAE,MAAM,wBAAwB,CAAC;AACzD,OAAO,EAAE,gBAAgB,EAAE,MAAM,yBAAyB,CAAC;AAC3D,OAAO,EAAE,eAAe,EAAE,MAAM,qBAAqB,CAAC;AACtD,OAAO,EAAE,iBAAiB,EAAE,MAAM,0BAA0B,CAAC;AAE7D,UAAU;AACV,OAAO,EAAE,sBAAsB,EAAE,YAAY,EAAE,iBAAiB,EAAE,MAAM,cAAc,CAAC;AAGvF,wCAAwC;AACxC,OAAO,EAAE,eAAe,EAAE,aAAa,EAAE,MAAM,qBAAqB,CAAC;AAGrE,sDAAsD;AACtD,OAAO,EAAE,oBAAoB,EAAE,cAAc,EAAE,mBAAmB,EAAE,MAAM,qBAAqB,CAAC;AAGhG,OAAO;AACP,OAAO,EAAE,YAAY,EAAE,UAAU,EAAE,gBAAgB,EAAE,gBAAgB,EAAE,MAAM,WAAW,CAAC;AAEzF,uBAAuB;AACvB,OAAO,EAAE,gBAAgB,EAAE,SAAS,EAAE,MAAM,EAAE,cAAc,EAAE,MAAM,sBAAsB,CAAC;AAC3F,OAAO,EAAE,YAAY,EAAE,cAAc,EAAE,gBAAgB,EAAE,MAAM,mBAAmB,CAAC;AAEnF,YAAY;AACZ,OAAO,EAAE,WAAW,EAAE,WAAW,EAAE,QAAQ,EAAE,SAAS,EAAE,MAAM,YAAY,CAAC"}
+{"version":3,"file":"index.js","sourceRoot":"","sources":["../src/index.ts"],"names":[],"mappings":"AAAA;;;;;GAKG;AAEH,6DAA6D;AAC7D,cAAc,gBAAgB,CAAC;AAE/B,2CAA2C;AAC3C,cAAc,oBAAoB,CAAC;AAEnC,cAAc;AACd,OAAO,EAAE,iBAAiB,EAAE,eAAe,EAAE,YAAY,EAAE,MAAM,iBAAiB,CAAC;AACnF,OAAO,EAAE,SAAS,EAAE,MAAM,kBAAkB,CAAC;AAC7C,OAAO,EAAE,gBAAgB,EAAE,MAAM,mBAAmB,CAAC;AACrD,OAAO,EAAE,eAAe,EAAE,MAAM,eAAe,CAAC;AAChD,OAAO,EAAE,iBAAiB,EAAE,MAAM,oBAAoB,CAAC;AAEvD,2BAA2B;AAC3B,OAAO,EAAE,sBAAsB,EAAE,YAAY,EAAE,iBAAiB,EAAE,MAAM,mBAAmB,CAAC;AAE5F,OAAO,EAAE,eAAe,EAAE,aAAa,EAAE,MAAM,mBAAmB,CAAC"}

package/package.json

@@ -1,31 +1,15 @@
 {
   "name": "kern-lang",
-  "version": "1.0.0",
-  "description": "Kern — the LLM-native language. Swiss-engineered IR that transpiles to React, Next.js, Express, and React Native.",
+  "version": "2.0.0",
+  "description": "The language LLMs think in. Write one .kern file, ship 7 targets. 70% fewer tokens.",
   "type": "module",
-  "bin": {
-    "kern": "./dist/cli.js"
-  },
   "main": "./dist/index.js",
   "types": "./dist/index.d.ts",
   "exports": {
-    ".": "./dist/index.js",
-    "./parser": "./dist/parser.js",
-    "./config": "./dist/config.js",
-    "./metrics": "./dist/metrics.js",
-    "./context-export": "./dist/context-export.js",
-    "./transpiler": "./dist/transpiler.js",
-    "./transpiler-web": "./dist/transpiler-web.js",
-    "./transpiler-tailwind": "./dist/transpiler-tailwind.js",
-    "./transpiler-nextjs": "./dist/transpiler-nextjs.js",
-    "./transpiler-express": "./dist/transpiler-express.js",
-    "./transpiler-cli": "./dist/transpiler-cli.js",
-    "./transpiler-terminal": "./dist/transpiler-terminal.js",
-    "./draft-protocol": "./dist/draft-protocol.js"
+    ".": "./dist/index.js"
   },
   "files": [
-    "dist",
-    "kern.config.ts"
+    "dist"
   ],
   "keywords": [
     "kern",
@@ -44,21 +28,18 @@
     "url": "git+https://github.com/cukas/KERNlang.git"
   },
   "homepage": "https://github.com/cukas/KERNlang#readme",
-  "license": "AGPL-3.0-or-later",
-  "scripts": {
-    "build": "tsc",
-    "prepublishOnly": "npm run build",
-    "test": "node --experimental-vm-modules node_modules/.bin/jest --forceExit",
-    "fitness": "npm test"
-  },
-  "devDependencies": {
-    "@types/jest": "^29.5.0",
-    "@types/node": "^22.0.0",
-    "jest": "^29.7.0",
-    "ts-jest": "^29.3.0",
-    "typescript": "^5.7.0"
-  },
+  "license": "AGPL-3.0",
   "dependencies": {
-    "jiti": "^2.6.1"
+    "@kernlang/protocol": "2.0.0",
+    "@kernlang/react": "2.0.0",
+    "@kernlang/native": "2.0.0",
+    "@kernlang/express": "2.0.0",
+    "@kernlang/core": "2.0.0",
+    "@kernlang/terminal": "2.0.0",
+    "@kernlang/cli": "2.0.0",
+    "@kernlang/metrics": "2.0.0"
+  },
+  "scripts": {
+    "build": "tsc -b"
   }
-}
+}

package/README.md

@@ -1,304 +0,0 @@
-# Kern
-
-**Write one `.kern` file. Ship 7 targets. Save 70% tokens.**
-
-Kern is the language LLMs think in. A high-leverage authoring language that transpiles to production stacks you already use — and cuts AI token costs by 70% when used as a communication protocol between competing AI engines.
-
-### Why Kern exists
-
-LLMs are expensive. Every token costs money. When an AI generates a React component, it outputs 500+ tokens of boilerplate — imports, StyleSheet, JSX, types. When two AIs discuss an implementation, they burn thousands of tokens on verbose natural language.
-
-Kern solves both problems:
-
-| Use case | Without Kern | With Kern | Savings |
-|---|---|---|---|
-| Generate a dashboard UI | 500+ tokens (React/TS) | 40 tokens (.kern) | **92%** |
-| AI proposes an approach | 800+ tokens (natural language) | 80 tokens (draft block) | **90%** |
-| 3 AIs brainstorm | 2,400+ tokens | 320 tokens | **87%** |
-| Express API with 4 routes | 400+ tokens | 40 tokens | **90%** |
-
-**For developers:** Write less, ship to more targets.
-**For AI systems:** Communicate in structured blocks instead of verbose prose.
-**For your wallet:** 70-90% fewer tokens = 70-90% lower API costs.
-
-```
-screen name=Dashboard {bg:#F8F9FA}
-  row {p:16,jc:sb,ai:center}
-    text value=FITVT {fs:24,fw:bold}
-    image src=avatar {w:40,h:40,br:20}
-  card {p:16,br:12,bg:#FFF,m:16}
-    progress label=Calories current=1840 target=2200 color=#FF6B6B
-    progress label=Protein current=96 target=140 color=#4ECDC4
-  button text="Log Meal" {w:full,br:8,bg:#007AFF,p:16}
-```
-
-That's 10 lines. Kern turns it into a complete React component with Tailwind classes, source maps, and token efficiency metrics. Or a Next.js page. Or a React Native screen. Or an Express API. Or a CLI. Or a terminal UI. Same source, 7 targets.
-
-## Targets
-
-| Target | Command | Generates |
-|---|---|---|
-| **Next.js** | `kern file.kern --target=nextjs` | App Router pages with metadata, `next/link`, `next/image` |
-| **Tailwind** | `kern file.kern --target=tailwind` | React + Tailwind CSS with `useState`, `useTranslation` |
-| **Web** | `kern file.kern --target=web` | React with inline CSS styles |
-| **React Native** | `kern file.kern --target=native` | React Native with `StyleSheet.create()` |
-| **Express** | `kern file.kern --target=express` | Typed Express routes with SSE streaming, child process spawn, timeouts |
-| **CLI** | `kern file.kern --target=cli` | Commander.js with typed args, flags, `parseAsync()` |
-| **Terminal** | `kern file.kern --target=terminal` | ANSI terminal UI — tables, spinners, progress bars, gradients |
-
-## Install
-
-```bash
-npm install kern-lang
-```
-
-Global install for CLI usage:
-
-```bash
-npm install -g kern-lang
-```
-
-## Quick Start
-
-```bash
-# Transpile to Next.js
-kern dashboard.kern --target=nextjs
-
-# Transpile to Express API
-kern api.kern --target=express
-
-# Transpile to CLI app
-kern mycli.kern --target=cli
-
-# Transpile to terminal UI
-kern ui.kern --target=terminal
-
-# Show language metrics
-kern dashboard.kern --metrics
-```
-
-Or without global install:
-
-```bash
-npx kern dashboard.kern --target=nextjs
-```
-
-## Express Backend Example
-
-```
-server name=API port=3001
-  middleware name=cors
-  middleware name=json
-
-  route method=post path=/api/review
-    schema body="{diff: string}"
-    stream
-      handler <<<
-        const results = await analyze(req.body.diff);
-        for (const r of results) emit(r);
-      >>>
-
-  route method=get path=/health
-    handler <<<
-      res.json({ ok: true });
-    >>>
-```
-
-Generates typed Express routes with SSE streaming, `AbortController` lifecycle, heartbeat keep-alive, and schema validation. Multi-file output via `GeneratedArtifact[]`.
-
-## CLI Example
-
-```
-cli name=mytool version=1.0.0 description="My CLI tool"
-  command name=build
-    arg name=target type=string required=true
-    flag name=watch alias=w type=boolean description="Watch mode"
-    flag name=timeout type=number default=30
-    import from="./build.js" names=runBuild
-    handler <<<
-      await runBuild(target, opts);
-    >>>
-```
-
-Generates Commander.js with `parseAsync()`, `parseFloat` coercion for number flags, `requiredOption()` for required flags, per-command files.
-
-## Terminal UI Example
-
-```
-screen name=Dashboard
-  gradient text="MY APP" colors=[208,214,220,226,228]
-  separator width=48
-  scoreboard title="Results" winner="engine-1"
-    metric name=Score values=["89","74","71"]
-    metric name=Time values=["45s","52s","38s"]
-  spinner message="Processing..." color=214
-  progress value=75 max=100 color=214
-```
-
-Generates pure Node.js ANSI escape codes — no dependencies. Tables, spinners, progress bars, gradients, boxes.
-
-## Metrics
-
-```bash
-npx kern examples/dashboard.kern --metrics
-```
-
-```
-Metrics: examples/dashboard.kern
-  Nodes:        23 (10 types)
-  Styles:       18 declarations
-  Mapped:       18 (100%)
-  Escaped:      0 (0%)
-  Shorthand:    94% coverage
-  Theme refs:   4
-```
-
-The metrics engine tells you exactly how much of your design system Kern handles natively vs. needs escape hatches.
-
-## Configuration
-
-```typescript
-// kern.config.ts
-const config: KernConfig = {
-  target: 'nextjs',
-  i18n: { enabled: true, hookName: 'useTranslation' },
-  components: { uiLibrary: '@components/ui' },
-  colors: {
-    '#18181b': 'zinc-900',
-    '#f97316': 'orange-500',
-    // your design system colors
-  },
-};
-```
-
-Config loaded via `jiti` (same as Tailwind CSS, Nuxt). CLI flags override config values.
-
-## API
-
-```typescript
-import {
-  parse,
-  transpileTailwind,
-  transpileNextjs,
-  transpileExpress,
-  transpileCliApp,
-  transpileTerminal,
-  collectLanguageMetrics,
-  resolveConfig,
-} from 'kern-lang';
-
-const ast = parse(kernSource);
-const result = transpileTailwind(ast, resolveConfig({ colors: myColors }));
-console.log(result.code);
-```
-
-## How It Was Built
-
-Kern was designed by three AI architectures — Claude (Anthropic), Codex (OpenAI), and Gemini (Google) — through competitive forge, brainstorm, and tribunal processes. Each feature was:
-
-1. **Brainstormed** — all 3 AIs propose approaches in Kern draft format
-2. **Forged** — all 3 implement independently, scored by automated fitness tests
-3. **Reviewed** — losing AIs critique the winner, bugs are fixed
-
-The Express target was forged in a 3-way competition. Codex won with typed generics and schema validation. Claude and Gemini's review caught 5 additional bugs. Every review found real issues — 9 review passes, 9 bugs caught and fixed.
-
-94 tests across 6 test suites. Zero type errors. Every commit verified.
-
-## Draft Protocol
-
-Kern isn't just a transpiler — it's a communication protocol between AI engines. The Draft Protocol lets competing AIs exchange structured proposals:
-
-```
-draft {
-  approach: "Use middleware chain with JWT validation"
-  reasoning: "Standard pattern, battle-tested"
-  tradeoffs: "adds latency", "requires secret management"
-  confidence: 82
-  keyFiles: "src/auth.ts", "src/middleware.ts"
-  steps {
-    1: "Add jsonwebtoken dependency"
-    2: "Create verifyToken middleware"
-    3: "Wire into Express app.use()"
-  }
-}
-```
-
-70% fewer tokens than natural language. Structured. Rankable. Engines speak Kern.
-
-## Token Savings — The Real Story
-
-Kern was born from a real problem: running 3 AI engines in parallel is expensive. Every brainstorm session, every forge competition, every code review burns tokens. At scale, this is the #1 cost driver for multi-AI systems.
-
-**Before Kern (natural language):**
-```
-I think we should implement this using a middleware chain pattern with JWT
-validation. The reasoning is that this is a standard, battle-tested approach
-that works well with Express. The main tradeoffs are that it adds latency
-per request and requires proper secret management. I'm about 82% confident
-this is the right approach. The key files we'd need to modify are
-src/auth.ts and src/middleware.ts. Here are the steps: first, add the
-jsonwebtoken dependency, then create a verifyToken middleware function,
-and finally wire it into the Express app.use() chain.
-```
-**~120 tokens**
-
-**After Kern (draft block):**
-```
-draft {
-  approach: "Use middleware chain with JWT validation"
-  reasoning: "Standard pattern, battle-tested"
-  tradeoffs: "adds latency", "requires secret management"
-  confidence: 82
-  keyFiles: "src/auth.ts", "src/middleware.ts"
-  steps {
-    1: "Add jsonwebtoken dependency"
-    2: "Create verifyToken middleware"
-    3: "Wire into Express app.use()"
-  }
-}
-```
-**~40 tokens** — same information, structured, parseable, rankable.
-
-**Real-world numbers from Agon (competitive AI orchestration):**
-
-| Operation | Old (natural language) | New (Kern) | Saved |
-|---|---|---|---|
-| 3-engine brainstorm | 2,400 tokens | 320 tokens | $0.02/call |
-| Forge plan review | 1,600 tokens | 240 tokens | $0.01/call |
-| Context injection | 800 tokens | 200 tokens | $0.005/call |
-| **Daily (50 operations)** | **~240K tokens** | **~38K tokens** | **~$2/day** |
-| **Monthly** | **~7.2M tokens** | **~1.1M tokens** | **~$60/month** |
-
-For a solo dev running multi-AI tools, that's the difference between affordable and not. For a company running thousands of operations, it's the difference between viable and bankrupt.
-
-Kern isn't just a language. It's a compression algorithm for AI communication.
-
-## License
-
-**AGPL-3.0-or-later** — You are free to use, modify, and distribute Kern under the terms of the [GNU Affero General Public License v3.0](LICENSE). This means:
-
-- If you modify Kern or build software that incorporates it, and you distribute that software or make it available over a network, you must release your source code under the same AGPL-3.0 terms.
-- This applies equally to individuals, companies, and organizations.
-
-**Commercial license:** If you want to use Kern in proprietary/closed-source software without the AGPL's source-sharing requirements, a commercial license is available. Contact [cukas](https://github.com/cukas) for details.
-
-**In practice:** Solo developers, students, and open-source projects can use Kern freely. Companies building closed-source products need a commercial license — or they can open-source their code.
-
-See [LICENSE](LICENSE) for the full text.
-
----
-
-## Contributors
-
-| Role | Who |
-|---|---|
-| **Creator & Director** | [cukas](https://github.com/cukas) |
-| **Co-Architect** | Claude (Anthropic) — Opus 4.6 |
-| **Co-Architect** | Codex (OpenAI) — GPT-5.4 |
-| **Co-Architect** | Gemini (Google) |
-
-Every feature was brainstormed by all 3 AIs, forged competitively, and cross-reviewed. 9 review passes, 9 bugs caught. The Express target was won by Codex in a 3-way forge. Engines speak Kern.
-
----
-
-**Swiss-engineered. AI-designed. Human-directed.**

package/dist/cli.d.ts

@@ -1,2 +0,0 @@
-#!/usr/bin/env node
-export {};