Private by default
No account. Learner data lives in a local SQLite database. AI off is a complete mode, not a degraded one.
Local-first adaptive learning
A personal tutor that runs on your machine.
Bring your own notes, textbooks, and labs. Cairn builds a skill map, adapts the next activity, and shows you why. Your data stays on your device. AI is optional.
macOS preview. Build from source for Windows or Linux. MIT licensed.
Git & GitHub Fundamentals
local
Practice · port mapping
Which flag publishes container port 80 on host port 8080?
Correct. The host port comes first.
Illustrative interface. Not a live session.
Why Cairn
Adaptation you can own and inspect.
Most adaptive platforms own the path, the content, and the data. Cairn takes the other position.
Your material
Import Markdown, HTML, DOCX, EPUB, and PDF. Bring Docker labs, lecture notes, and textbooks. Build and export packs.
Visible adaptation
Every activity has a reason. Change difficulty, help, AI, or path. Switch to manual mode when you want the wheel.
Packs
Start with a real subject.
Packs include explainers, practice, retention review, and learner-run labs.
Docker Fundamentals
Containers, images, ports, volumes, networks, Dockerfiles, Compose, and troubleshooting.
Git & GitHub Fundamentals
Commits, branches, merges, conflicts, remotes, pull requests, ignores, and recovery.
A community library is planned after the pilot. The local library is always the source of truth.
How it works
Import, practise, see why, steer.
01
Bring material
Import notes, textbooks, or a lab repository.
02
Start a session
Cairn diagnoses, then chooses an activity and explains why.
03
Adjust
Change difficulty, help, AI, or switch to manual mode.
04
Review
Retention reviews return when they are useful, with a cap.
Trust
Your learning data is not the product.
No account, no learner database in the cloud, no telemetry by default. Export everything. Delete everything. Turn AI off.
Status
Pilot-ready, not finished.
Cairn v0.2 is a research-grade prototype. The adaptive model, retention controls, packs, authoring, and local export are working. A public pilot and a signed release are next. The adaptive methods are inherited from established work. The local-first, learner-owned workflow is the contribution.