junjiezhou1122

OpenHyperKnow — AI skill for Claude Code

AI community

Open-source AI learning platform — Hyperknow clone built on pi AgentSession SDK.

How to install OpenHyperKnow

This entry records only its repository, not the path inside it, so there is no exact command to give. Open junjiezhou1122/OpenHyperKnow and copy the folder into ~/.claude/skills/, or the file into ~/.claude/agents/.

What OpenHyperKnow does

Open-source AI learning platform — Hyperknow clone built on pi AgentSession SDK. Course generation, adaptive learning, assessments.

Alternatives in AI

WeKnora — Open-source LLM knowledge platform: turn raw documents into a queryable RAG, an autonomous reasoning agent, an 26.9k ★
Osaurus — Own your AI. The native macOS harness for AI agents -- any model, persistent memory, autonomous execution, cry 5.1k ★
Axonhub — ⚡️ Open-source AI Gateway — Use any SDK to call 100+ LLMs 3.2k ★

README

OpenHyperKnow

Open-source AI learning platform — a faithful clone of [Hyperknow](https://hyperknow.io) built on **[pi](https://github.com/badlogic/pi-mono)**'s AgentSession SDK instead of LangChain.

Turn any learning goal into a 1:1 AI course: web research → profile questions → adaptive course → lessons → practice → assessments.

Features

🎯 Course Generation — enter a topic, the agent researches university syllabi, asks 4 profile questions (prior knowledge / lens / mastery / scale), then streams a full course unit-by-unit
📚 Adaptive titles — course title/description adapt to your profile answers
📖 Course reader — unit → lecture → lesson navigation, markdown + KaTeX rendering
✍️ Practice — Q/A reveal cards per lesson
🏆 Assessments — MCQ per unit with instant scoring (Mastered / Proficient / Review)
🔌 Multi-provider LLM — pick any provider/model pi supports (GitHub Copilot, MiniMax, Codex, Anthropic, OpenAI, Ollama, ...) — nothing is hardcoded

Whiteboard (interactive teaching)

The 🖊️ Whiteboard tab reproduces Hyperknow's teaching session:

The agent narrates (left stream) while placing handwritten note cards (Kalam font, cream paper) and mermaid diagrams on a dotted-grid canvas
Multi-page lectures with ‹ › navigation
Interrupt: the agent pauses mid-lecture to ask a question; the student answers; the agent acknowledges and continues from that point
Every element carries a step_id (Hyperknow's sync protocol)

Architecture

OpenHyperKnow/
├── backend/        Node.js + Fastify + WebSocket + pi AgentSession SDK
│   └── src/agents/teacher.ts   ← replaces LangGraph: tools + interrupt + streaming
└── frontend/       React 19 + Vite + Tailwind v4
    └── src/hooks/useCourseGenSocket.ts  ← Hyperknow-style WS protocol + reducer

The agent workflow is **not** a framework graph — it's a pi `AgentSession` with 4 custom tools: