How do I install the Learn skill?

Run npx skillmds@latest add lemonaderrrrr/learn in your terminal (requires Node.js), paste this page's agent-chat prompt into Claude, Cursor, or any MCP-connected agent, or download the SKILL.md file and copy it into your agent's skills directory.

Is Learn safe to use?

This skill has not completed SkillMD's automated safety review yet. SkillMD never runs a skill's scripts for you; review the SKILL.md before installing.

Which AI agents work with Learn?

This skill is tagged as working with Claude Code, Claude.ai, OpenAI Codex. SKILL.md is an open format, so most agents that read a skills directory can load it too.

Is Learn free to use?

Yes. Installing skills from SkillMD is free, and the skill stays under its author's original license.

Lemonaderrrrr (@lemonaderrrrr) published this skill. Their other Agent Skills are listed on their SkillMD profile.

Learn by lemonaderrrrr — Agent Skill

Name: Learn
Author: Lemonaderrrrr

Learn

Absorb a topic, paper, or lesson into reusable knowledge. Default Philosopher mode (first-principles decomposition + optional Feynman check + a distilled note); Engineer mode for time-boxed extraction of usable takeaways. Saves notes to outputs/notes/. Use when the user says "I want to learn X", "help me understand this paper", or invokes the skill.

File contents

Learn Workflow

Turn a topic / paper / lesson into reusable, durable knowledge.

Start

Route-confirm (see CONVENTIONS.md): in one line, say this is the learn workflow and which mode you suggest; wait for the go-ahead.
Confirm mode: default 🧠 Philosopher; the user may switch to ⚙️ Engineer.
(Optional) Anchor the material to the user's own focus area so new knowledge connects to what they care about.

🧠 Philosopher mode (default)

Goal: truly understand, reusable for life. Slow is fine; half-baked is not.

Frame — What is this? What field? Why does it matter?
Decompose to first principles — Break down to irreducible concepts; at each layer ask "why this way, why not otherwise?"
Feynman check (optional) — Have the user explain it back in plain words. Where they stumble = not yet understood → go back and fill the gap. Skippable on request.
Connect — Link to what they already know (prior topics, notes).
Capture — Write outputs/notes/note_<english-slug>.md: one-sentence essence + 3 key insights + open questions.

⚙️ Engineer mode

Goal: get usable conclusions, time-boxed. For exams / reproducing a result / looking up one thing.

Anchor the goal — What will you do with this? (exam / reproduce / ship)
Extract — Go straight to takeaways + key formulas / APIs; skip derivations.
Action list — Turn it into executable steps or a code skeleton.
Capture — Write outputs/notes/cheat_<english-slug>.md: a quick-reference card.

Wrap-up conventions

Live-search principle: look up external facts (papers, recipes, APIs, data) online; don't rely on training memory. Pure reasoning / math is exempt.
Promote: if it's directional/reusable, offer to save it to long-term memory; one-off outputs stay put.
Filenames use English slugs (e.g. note_transformers.md).
Follow the host repo's CLAUDE.md style if present.

---name: learndescription: Absorb a topic, paper, or lesson into reusable knowledge. Default Philosopher mode (first-principles decomposition + optional Feynman check + a distilled note); Engineer mode for time-boxed extraction of usable takeaways. Saves notes to outputs/notes/. Use when the user says "I want to learn X", "help me understand this paper", or invokes the skill.---# Learn WorkflowTurn a topic / paper / lesson into **reusable, durable knowledge**.## Start1. **Route-confirm** (see `CONVENTIONS.md`): in one line, say this is the learn workflow and which mode you suggest; wait for the go-ahead.2. **Confirm mode**: default 🧠 **Philosopher**; the user may switch to ⚙️ Engineer.3. *(Optional)* Anchor the material to the user's own focus area so new knowledge connects to what they care about.## 🧠 Philosopher mode (default)Goal: **truly understand, reusable for life**. Slow is fine; half-baked is not.1. **Frame** — What is this? What field? Why does it matter?2. **Decompose to first principles** — Break down to irreducible concepts; at each layer ask "why this way, why not otherwise?"3. **Feynman check (optional)** — Have the user explain it back in plain words. Where they stumble = not yet understood → go back and fill the gap. Skippable on request.4. **Connect** — Link to what they already know (prior topics, notes).5. **Capture** — Write `outputs/notes/note_<english-slug>.md`: one-sentence essence + 3 key insights + open questions.## ⚙️ Engineer modeGoal: **get usable conclusions, time-boxed**. For exams / reproducing a result / looking up one thing.1. **Anchor the goal** — What will you do with this? (exam / reproduce / ship)2. **Extract** — Go straight to takeaways + key formulas / APIs; skip derivations.3. **Action list** — Turn it into executable steps or a code skeleton.4. **Capture** — Write `outputs/notes/cheat_<english-slug>.md`: a quick-reference card.## Wrap-up conventions- **Live-search principle**: look up external facts (papers, recipes, APIs, data) online; don't rely on training memory. Pure reasoning / math is exempt.- **Promote**: if it's directional/reusable, offer to save it to long-term memory; one-off outputs stay put.- Filenames use **English slugs** (e.g. `note_transformers.md`).- Follow the host repo's `CLAUDE.md` style if present.

Frequently asked questions