Not a chat wrapper. Not another tutorial site.
Messier turns any developer topic into connected lessons, active recall, spaced repetition, and real practice so you actually retain what you learn.
Every lesson follows the same proven structure — designed around how developers actually learn, not how they consume content.
Clear, dense explanations. Just the mental model you need.
ReadA read-only working example. Watch it come alive before you touch anything.
ObserveWrite real code. Blank slate or fill-in-the-blank. Get instant feedback.
ChallengeAI tutor on-demand. Learning first, assistance second.
On-demandEvery decision in Messier pushes against passive learning. This is what that looks like under the hood.
Bring your own Anthropic API key. We never store it — it lives in your browser. You pay only for what you use, at cost. No markup, no lock-in. A month of serious learning costs less than a single textbook.
The SM-2 algorithm schedules reviews at the exact moment you're about to forget. Active recall, not passive re-reading.
A live D3.js graph maps every concept you've learned and how they connect. Watch your knowledge compound over time.
Messier turns completed lessons into Zettelkasten-style atomic notes, links them to your knowledge graph, and schedules them into review so ideas do not disappear after the lesson ends.
Upload a PDF, paste notes, or drop in documentation. Messier builds lessons, challenges, and review prompts around the source instead of generic internet-shaped content.
PDFs, markdown, code files, transcripts, or pasted notes.
The curriculum stays grounded in the material you provided.
Challenges and spaced reviews reinforce the source concepts over time.
Cognitive science has known for over a century how humans retain information. Most learning tools ignore it. Messier doesn't.
Hermann Ebbinghaus mapped human memory decay in 1885. Without reinforcement, we forget most of what we learn within a day. The curve is steep and unforgiving.
Spaced repetition — reviewing material at increasing intervals — flattens that curve. Each review resets and extends retention. After enough repetitions, knowledge moves from working memory into long-term storage.
The SM-2 algorithm, originally developed for SuperMemo and popularised by Anki, calculates the optimal moment to review each concept — not too early, not too late.
Wozniak, P.A. (1990). Optimization of learning. Unpublished MSc thesis, University of Technology, Poznań.Retrieval practice produces greater gains in long-term retention than elaborative studying with concept mapping.
Distributing practice across time produces better long-term retention than massing practice in a single session.
Whether you want full control over your stack or just want to start learning — Messier works both ways.
Zero setup. Connect your Anthropic API key and start learning in under a minute. We handle auth, storage, spaced repetition, and the knowledge graph.
Run Messier entirely on your own infrastructure. Bring your own database, AI provider, and code runner. MIT licensed — fork it, extend it, own it.
Not sure? Start with Cloud — you can always self-host later.
Jump into courses or open your dashboard.
Pick a course type and your usage pattern. See exactly what to expect in API credits — before you commit to anything.
Based on Claude Sonnet pricing (~$3/M input, ~$15/M output).
You pay Anthropic directly — Messier takes nothing from your API usage.
Charles Messier (1730–1817) was a French astronomer who spent his career hunting comets. Along the way he kept mistaking nebulae and star clusters for the comets he sought — so he compiled a catalog of 110 objects just to remember what wasn't a comet.
That catalog — the Messier Catalog — became one of the most enduring documents in astronomical history. He was trying to find something else, and he ended up mapping the deep sky.
This app works the same way: you come to learn a topic, and along the way you build a catalog of ideas that stays with you long after the lesson ends.