LearnKit AI

The open-source AI workbench for teams that ship. Generate personalized, role-aware learning paths for Claude, Cursor, ChatGPT and 40+ tools - and embed them in your product in 3 lines.

Most enterprise AI training is a video library nobody finishes. LearnKit AI is project-based: every lesson ends with a real prompt, agent, or workflow your team can ship that Friday - reviewed by the AI Guide.

pnpm add @learnkit-ai/core @learnkit-ai/react

import { LearningPath } from '@learnkit-ai/react';

<LearningPath
  input={{
    role: 'Product Manager',
    tools: ['Claude', 'Cursor'],
    goal: 'Ship an internal research agent',
    level: 'beginner',
  }}
/>;

That renders a four-week, twelve-lesson curriculum tuned to that role + stack + goal. No backend, no LLM call, no API key. The engine is deterministic - the same input always produces the same path.

---

Why LearnKit AI

Most AI training	LearnKit AI
10-hour video libraries	30-day, project-based paths
Completion certificates	A portfolio of shipped projects
Generic content for everyone	Adaptive to your role, stack, and goal
Closed platform	Apache-2.0 source, embeddable in your product
Stale within a quarter	Tools tracked, updated every two weeks

Features

• generateLearningPath() - pure function, no network, fully deterministic
• <LearningPath /> - drop-in React component with three built-in themes (warm, midnight, technical)
• <LessonCard /> - primitive for custom layouts
• <AIGuide /> - embeddable AI tutor avatar + tip card
• useLearnKit(input) - React hook for headless use
• Zod schemas for everything - strict input validation and inferred types
• No Tailwind dependency in @learnkit-ai/react - CSS custom properties only, drops into any host theme
• Apache-2.0 - fork it, ship it

Quick start

1. Use the engine standalone

import { generateLearningPath } from '@learnkit-ai/core';

const path = generateLearningPath({
  role: 'Software Engineer',
  tools: ['Cursor', 'Claude'],
  goal: 'Ship a feature without bothering anyone on the team',
  level: 'intermediate',
});

console.log(path.weeks.length);                // 4
console.log(path.weeks[0].lessons[0].title);   // "Your first system prompt"
console.log(path.totalMinutes);                // ~430

2. Render the path in any React app

import { LearningPath } from '@learnkit-ai/react';

export default function MyPage() {
  return (
    <LearningPath
      input={{
        role: 'Designer',
        tools: ['Midjourney', 'ChatGPT'],
        goal: 'Generate a brand-faithful image library',
        level: 'beginner',
      }}
      theme="midnight"
      onLessonClick={(lesson) => console.log(lesson)}
    />
  );
}

3. Headless with the hook

import { useLearnKit } from '@learnkit-ai/react';

function MyCustomUI({ input }) {
  const { path, error } = useLearnKit(input);
  if (error) return <Error error={error} />;
  if (!path) return null;
  return <pre>{JSON.stringify(path, null, 2)}</pre>;
}

4. Drop in the AI Guide

import { AIGuide } from '@learnkit-ai/react';

<AIGuide
  message="Looks like you're about to write a system prompt. Want a 2-minute refresher?"
  animated
/>;

Packages

This is a pnpm + Turborepo monorepo.

Package	Description
@learnkit-ai/schemas	Zod schemas + inferred TypeScript types. No logic.
@learnkit-ai/core	generateLearningPath() and role/tool helpers. Pure, no network.
@learnkit-ai/react	React components: <LearningPath />, <LessonCard />, <AIGuide />.
apps/web	Marketing site, demo, docs at learnkit-ai.com.
examples/nextjs-basic	Minimal Next.js integration.

schemas ← core ← react ← apps/web

Public API

// @learnkit-ai/core
function generateLearningPath(input: LearningPathInput): LearningPath;
function getSupportedRoles(): string[];
function getSupportedTools(): string[];
function isRoleSupported(role: string): boolean;

// @learnkit-ai/schemas
type LearningPathInput = {
  role: string;
  tools: string[];
  goal: string;
  level: 'beginner' | 'intermediate' | 'advanced';
  companyContext?: string; // max 500 chars
};

// @learnkit-ai/react
<LearningPath input={LearningPathInput} onLessonClick? theme? />
<LessonCard lesson={Lesson} status? onClick? />
<AIGuide message={string} animated? size? />
useLearnKit(input): { path, loading, error }

The LearningPathInput shape is the public API contract. It will not change without a major version bump.

Development

Requirements: Node ≥ 20, pnpm ≥ 9.

git clone https://github.com/learnkit-ai/learnkit.git
cd learnkit
pnpm install

pnpm dev          # apps/web at localhost:3000
pnpm test         # vitest across all packages
pnpm typecheck    # tsc across all packages
pnpm build        # build all packages and apps
pnpm lint         # lint all workspaces

How to contribute

Issues, PRs, and discussions are welcome. Some great first contributions:

• Add a tool - pick a tool from the tracker issue, open a PR to packages/core/src/tools.ts with a curriculum template
• Improve a lesson template in packages/core/src/generate.ts
• Add a theme to <LearningPath /> in packages/react/src/LearningPath.tsx
• Write a blog post in apps/web/src/lib/blog-posts.ts
• Translate the UI - locales are wide open

See CONTRIBUTING.md for guidelines and CODE_OF_CONDUCT.md for community expectations. Security issues: please follow SECURITY.md.

Roadmap

• v0: schemas, core, react packages, marketing site
• v0.2: SSR-safe streaming for <AIGuide />
• v0.3: Eval rubrics in @learnkit-ai/evals
• v0.4: 40-tool curriculum library (currently 8)
• v1: stable public API, full docs site, deployment recipes

Nothing on the roadmap will be paid-only. The license is Apache-2.0 for the long haul - fork it, ship it, run it without us.

License

Apache-2.0. Use it, fork it, ship it - including commercially.

Acknowledgements

LearnKit AI is built on top of years of work by the open-source community. Particular thanks to the maintainers of Zod, Next.js, React, and TypeScript.

---

LearnKit AI · learnkit-ai.com · @learnkit-ai