Search for a command to run...
Explore catalog courses with the AI tools you already use. Highlight, summarize, and customize lessons — rooted in reputable sources, connected to Claude, Codex, Cursor, and Gemini through MCP.
Free for learners. Free for teachers. No credit card.
Read. Connect. Make it yours.
The Model Context Protocol (MCP) is an that lets LLMs interact seamlessly with external tools, APIs, and data sources.
MCP follows a — MCP Hosts run the AI models, MCP Clients initiate requests, and MCP Servers serve context, tools, and capabilities.
It acts as a , much like USB-C standardized physical connections for devices.
Trapped in one publisher's app, one LMS, one URL.
Textbook + courseware bundles north of $300 per class.
Print cycles can't keep up with the field.
Bolted-on chatbots that don't know what you're reading.
Teachers can't customize, learners can't reshape.
Choose courses from the catalog and study them in LearnOS or through MCP with the AI tools you already use — Claude, Codex, Cursor, Gemini — to read, summarize, quiz, and save notes on your terms.
Study catalog lessons with searchable sections, an AI tutor, and citations that take you back to the source.
github.com/stewart/calc-textbookEvery lesson renders as long-form prose — KaTeX-formatted math, Mermaid diagrams, syntax-highlighted code. Highlight, take notes, and chat with the lesson without leaving the page.
Select any passage. Highlights persist per-user and feed quick actions (summarize · flashcard · ask) right on the spot.
Every assistant message resolves cite:LESSON#ANCHOR markers into clickable chips that open the exact heading section.
One-click comprehensive audio overview of a lesson or module — a single AI narrator covers the whole arc. Streams while you read, or queues up for the gym.
Generate SM-2-scheduled decks from a lesson, a highlight, or a custom prompt — reviewed in /today.
Hover the badges on the screenshot to see each piece in detail.
For a function f, the derivative at a point a is defined as the limit of (f(a+h) − f(a)) / h as h approaches zero.
Geometrically, this is the slope of the tangent line at a — the best linear approximation to the curve at that point.
When the limit exists, we say f is differentiable at a.
Choose a catalog course and add it to your library. Search by keyword or meaning, then open the exact section you need. Course imports are managed by admins.
A single AI narrator walks through the entire lesson — citation-aware, comprehensive enough to stand on its own. Generated server-side, streamed in the background while you read, queued up for the gym.
Scope a question to one lesson, a module, or your whole library. Hybrid retrieval (pgvector + tsvector RRF) ranks the relevant sections; every answer cites the exact heading anchor it pulled from.
Select any passage, pick the Flashcard action, and a card lands in your deck — scheduled with SM-2, queued in /today. Generate a whole deck from a lesson in one tap.
A function is continuous at a point when the limit equals the value.
Everything you need to read, retain, and apply — built for serious self-directed learners. No paywalls. No vendor lock-in.
Generate a comprehensive audio overview of any lesson or module — one AI narrator, citation-aware, covers the full arc. Streams in the background while you read or commute.
Ask a question scoped to a lesson, a module, a course, or your whole library. Every answer cites the exact heading section it pulled from.
One-tap card generation from a lesson or highlight, scheduled with SM-2-lite. Review the queue in /today with cards bundled by course.
Markdown in. Citations checked. Quality scored. Push to git and your students see the update on next page load.
Course authors can prepare markdown lessons with their preferred tools. Admins manage imports into LearnOS; learners study courses from the catalog.
---
title: The derivative as a limit
module: 02-calculus-foundations
lesson: 04-derivative-limit
prereqs: [limits, function-notation]
objectives:
- Define the derivative as a limit
- Connect to tangent line geometry
---Claude Code, Codex, Gemini CLI, Cursor — point any of them at your course repo and let them draft modules, run a citation-verifier pass, regenerate flashcards.
Generate a structured quality report on any course: source citation density, prerequisite chain integrity, exercise coverage. Publish gates use the rubric so the catalog stays high-signal.
LearnOS exposes search, lessons, transcripts, flashcards, and notes over MCP. Drop a token into any compatible agent and your courses are right there.
The textbook hasn’t changed in fifty years. The LMS hasn’t changed in twenty. The AI tools that are changing — Claude, Codex, Cursor — can’t see what you’re trying to learn.
LearnOS is the missing layer: a catalog of durable courses, with AI retrieval and MCP integration as table stakes — not a $20/month upsell.
We don't host content. We point at the best material that already exists — textbooks, lectures, open-source curricula — and add the layer that makes it studyable, citable, and yours.
Chat, summaries, flashcards, citation-grounded retrieval — all on by default, everywhere, for every learner. Not a Pro tier. Not a chatbot in a corner.
MCP-first. Your courses belong inside Claude, Codex, Cursor, Gemini — wherever you already work. We don't compete with the AI tools you love; we make them better.
Ask about a lesson, share what you made, and meet learners working through the same ideas. LearnOS is the workspace; the community is where the conversation continues.
Sign in once. Bring your AI tool of choice. Free for learners, free for teachers.
Three ways to make an AI answer from your own notes and courses and cite the exact source, compared on citations, where your material lives and which AI you can use.
An agent harness is the code around a model that turns it into an agent. Here is what it does, and the evidence that it can matter as much as the model.
The top coding-agent harnesses compared on the decisions that matter: tools, edit format, permissions, sessions, extensibility and which models they run.