@AIExplorerTim: Learners, check this out — these 4 skills let AI tutor you like a personal instructor. It doesn't just hand you answers; it keeps probing until you truly understand. And they're all free. 1. Socratic Tutor — Instead of giving answers directly, it keeps asking questions until you figure it out yourself. Works with any LLM. https…

X AI KOLs Timeline Tools

Summary

Introduces 4 free AI learning skills that make an LLM act like a personal tutor — guiding you to deep understanding through Socratic questioning, the Feynman technique, a learning log, and conversational memory. The core Socratic Tutor was open-sourced by Jane Street.

Learners, check this out These 4 skills let AI tutor you like a personal instructor Not just giving answers, but probing until you truly understand And they're all free 1. Socratic Tutor Instead of handing over answers, it keeps asking questions until you work things out for yourself — works with any LLM 🔗 https://github.com/janestreet/agent_skills… 2. Supertutor Combines the Feynman technique with Socratic questioning, teaching until you fully get it, catching you when you try to gloss over things — great for exam prep too 🔗 https://github.com/cskwork/supertutor-skill… 3. Learning Log Records what you've learned; AI reads the log to adjust how it teaches you — works with any AI 🔗 https://github.com/lukasreese/claude-skills-starter-kit/tree/main/skills/learning-log… 4. Session Memory Lets it remember past conversations so your learning never breaks off — pick up right where you left off yesterday 🔗 https://github.com/lukasreese/claude-skills-starter-kit/tree/main/skills/session-memory…
Original Article
View Cached Full Text

Cached at: 11/20/26, 04:07 PM

For Learners

These 4 skills will have AI tutor you like a personal coach

Not just giving answers — it keeps questioning until you truly understand. And all of them are free.

1. Socratic Tutor (苏格拉底式家教) Instead of giving answers directly, it keeps asking questions until you figure things out yourself. Works with any LLM. 🔗 https://github.com/janestreet/agent_skills…

2. Supertutor (精通式家教) Combines the Feynman technique with Socratic questioning, drilling until you fully understand. It calls you out when you’re just bluffing — also great for exam prep. 🔗 https://github.com/cskwork/supertutor-skill…

3. Learning Log (学习记录) Keeps a record of what you’ve learned. The AI reads this log and adapts how it teaches you. Works with any AI. 🔗 https://github.com/lukasreese/claude-skills-starter-kit/tree/main/skills/learning-log…

4. Session Memory (对话记忆) It remembers previous conversations so your learning never breaks. Pick up right where you left off yesterday — even review from that point directly. 🔗 https://github.com/lukasreese/claude-skills-starter-kit/tree/main/skills/session-memory…


janestreet/agent_skills

Source: https://github.com/janestreet/agent_skills

Agent skills from Jane Street

A collection of skills for AI agents and chatbots. Each skill is a directory containing a SKILL.md file with instructions for the agent.

Skills

  • socratic-tutor — turns an LLM into a strict personal tutor that teaches a topic of your choice using the Socratic method: a diagnostic, a skill map with levels 0-5, one question at a time, spaced review, and a final assessment.

How to use

With a coding agent that supports the common SKILL.md format, copy the skill directory into your agent’s skills folder, then ask the agent to use it (“teach me X”).

With a plain chatbot, paste the body of the skill’s SKILL.md into the chat (or into a custom/system prompt), followed by what you want it to do.

Contributing

These skills are exported from our internal repository, where we use them day to day, and we hope it’s interesting to share that experience more widely. Since we’re trying to share what we’ve found to be useful, we’re not looking for new skills or for pull requests that change the text of existing ones. If you’ve adapted a skill and it works better for you, that’s great, but please keep your own copy.

Issues are welcome if something doesn’t work as described, though we may not be able to act on all of them.

Prior art

Teaching by asking questions is the Socratic method (https://en.wikipedia.org/wiki/Socratic_method), which is rather older than LLMs, and there are other tutoring prompts and skills built on the same idea. Our skill was written internally and isn’t knowingly derived from any other implementation.

Similar Articles

@mylifcc: While almost every company is teaching employees how to use AI to generate ready-made answers with a single click, Jane Street, Wall Street’s most profitable quantitative trading giant and one with the most hardcore taste in technology, went the other way and open-sourced its internal janestreet/agent_skills repository on GitHub, which shot up …

X AI KOLs Timeline

Quantitative trading giant Jane Street has open-sourced its internal janestreet/agent_skills repository on GitHub. Its Socratic Tutor skill, which made it to the front page of HN, uses rigorous Socratic questioning and tiered assessment rules to deliberately create cognitive friction for learners, preventing the erosion of thinking skills that can result when AI provides answers directly.

@Aoyi21: https://x.com/Aoyi21/status/2064975015200829457

X AI KOLs Timeline

This article proposes that the most cost-effective way to learn AI is to deconstruct others' Skills. By analyzing their task definitions, trigger conditions, operation steps, prohibitions, and acceptance criteria, you can learn how experts think and train AI, rather than just using tools.

@realCaigu: A former Google CEO highlighted a key insight: If you really want to make money, you must learn to understand agents, Claude Code, prompts, memory, skills, MCP, and routines — then let your AI control your AI. Most people learn AI by scrolling through tweets, but the truly systematic knowledge is available for free from the major AI model companies. Be sure to bookmark this post.

X AI KOLs Timeline

A former Google CEO shares the key directions for learning AI (agents, Claude Code, prompts, etc.) and recommends a series of free official learning resources from LangChain and Anthropic.

@slashui: I’m increasingly convinced that textbooks need to be redone by AI! Just discovered another fantastic open-source project: Teachable Figures. It turns an AI agent into an interactive course designer that can generate beautiful 2D and 3D educational scenes. For example: Transformer…

X AI KOLs Timeline

Introducing the open-source project Teachable Figures, which turns an AI Agent into an interactive course designer capable of generating 2D/3D educational scenes (such as Transformer attention mechanisms, mechanical structures, and human anatomy). It supports active learning through prediction, hands-on exploration, and observation, works in both Chinese and English, and can run offline.