One command turns your reading
into an activity you can’t skim.
Type /explain, /quiz, or a Socratic /derive — nearly 40 commands, each turning your own material into a live activity.
Many are interactive; the AI grades what you write.
The core move
Type it. Do it. Get graded.
One command, one continuous flow — start to finish.
- Step 1
One word, on your own material.
- Step 2
Generated live — not a pre-made worksheet.
- Step 3
You can’t skim past it — you answer, then you find out.
- Step 4
Write your own answer — the AI grades what you wrote, not a checkbox.
That's one command. There are nearly 40 more — one for every way you need to think.
The gap
You’ve read it twice and had a chatbot explain it. It’s still not yours.
Trap 1
The fluency illusion
Re-reading and highlighting feel like progress. They build recognition — that warm “yeah, I know this” when you see the answer. They don't build recall, which is the only thing an exam or a real problem ever asks for. You close the book sure. The blank page disagrees.
Trap 2
A chatbot does your thinking for you
Paste a hard idea into a generic chatbot and it hands back a clean explanation. Efficient — except it did the retrieving, the connecting, the deciding, all the work that would have made it stick. You got an answer, not a skill. And nothing ever checked whether you could do it without the bot.
Learning runs on the opposite: pull it from memory, wrestle with it a little, get told exactly where you're wrong. That's all a command is — one word that flips your own material from something you read into something you do, then grades what you produce.
The library
One box. Nearly 40 ways to think.
Every command is one focused activity. They group by what you’re trying to do — pick a family.
- commands
- 45commands
- interactive
- 18interactive
- Socratic-guided
- 12Socratic-guided
- free response
- AI-gradedfree response
Understand it — meet a concept and make it click
10 commands5 guided
- /explaina clear, structured explanation
- /eli5explain it like you’re five
- /intuitionthe feel behind the formalism
- /analogymap it to something you know
- /whythe reason it’s true
- /howhow it works, step by step
- /examplea worked, concrete example
- /contextwhere it fits in the bigger picture
- /prereqwhat to learn first
- /summarythe whole thing, distilled
interactive activitySocratic guided mode
Practice commands like /quiz, /flashcard, and /test also take a --bloom flag — aim it at remember, apply, or create, and the activity changes with it.
Under the hood
You typed one word. Look what it was carrying.
You ran /quiz a moment ago. Here's everything that one word carried.
Layer 1
You can aim it
/quiz isn't one quiz. Add --bloom=apply and it tests whether you can use the idea, not just recall it. Almost every command takes flags like these: one word, or a precise instrument.
Layer 2
It’s built on learning science
The questions follow Bloom's revised taxonomy (Anderson & Krathwohl, 2001). /analogy obeys Cognitive Load Theory; /askme's feedback follows formative-assessment research. You never see the citations — you feel the difference.
Layer 3
The AI stays in the loop
When you write your own answer, the AI grades it 0–10, names the exact misconception, and hints — instead of just handing you the solution.
Layer 4
Generated live, from your material
No two runs are identical. Every activity is written on the spot from the page you're on — never pulled from a pre-made bank.
One word. Four layers of machinery. Times nearly forty.
Guided mode
It doesn't just quiz you. It leads you to the answer.
Every teaching command runs one of two ways. --mode=direct hands you a crisp answer. --mode=guided refuses to — it leads.
- Ritsu asks · rung 1A seed weighs a gram; a full oak weighs tons. Where did all that matter come from — the soil?
- You answer4/10From the soil, I'd guess — the roots pull it up?
- Ritsu asks · rung 2Reasonable — but the soil in that experiment barely loses weight. What does a leaf take in all day?
- You answer8/10CO₂, from the air.
- Ritsu asks · rung 3So the carbon in that CO₂… where does it end up?
- You answer10/10In the sugars it builds — the wood itself. The tree is made of air.
A tree is built mostly from air.
It climbs a ladder of 4–6 questions, scores each answer, names the exact place your reasoning slips, and hints when you stall — until you reach the insight, not Ritsu.
Say “I'm stuck” anytime and it catches you.
Twelve commands can go guided:
- /explain
- /why
- /how
- /intuition
- /analogy
- /derive
- /what-if
- /counter-example
- /contrast
- /compare
- /exercise
- /solve
That's the difference between being told and being taught.
The switch
Every other study tool starts empty. These start full.
- A flashcard app you fill in by hand
/flashcard - A quiz / test maker
/quiz/test - A tutor you pay by the hour
/explain --mode=guided - Problem sets from a workbook
/solve/exercise/code - A mind-mapping app
/mindmap/concepts - Summaries you write yourself
/summary
Every tool in that stack starts empty — you feed it. Every command here starts full — it already knows the page you're on.
One box. No tab-switching, no copy-paste, no setup. Just the next command.
Questions
What people ask before they type their first command.
No. Type / and the box lists them, grouped, with one-line descriptions. Most people live on five or six — start with /explain and /quiz, add more when you want them. /help shows everything.
Your first command is one keystroke away.
Bring in a PDF, a video, or your own notes. Type /explain. See what one word does.