Learn anything, adapt in real time.
SkillCoco is an open-source desktop app that replaces “did they watch the video?” with “can they actually do this?” — powered by real learning science, AI-generated content, and hands-on terminal labs you can build yourself.
Kubernetes Fundamentals
Module 4 · Pod scheduling
Mastery — Bayesian Knowledge Tracing
nginx-7d9f8 1/1 Running
✓ Step 3 check passed
What makes SkillCoco different
Six things most learning apps skip — none of them optional here.
Real Bayesian Knowledge Tracing
Corbett & Anderson's 1994 BKT algorithm — the same model behind Carnegie Mellon's Cognitive Tutors. Every answer updates a per-skill posterior; modules unlock only when mastery is demonstrated.
AI-judged terminal labs
Labs check work in three layers: regex, exit-code, and — as a last resort — a budget-guarded AI judge scoring open-ended steps against a per-step rubric. Evaluation never silently passes.
Spaced repetition that decays
Mastered modules auto-generate SM-2 flashcard reviews on Wozniak's forgetting-curve schedule. The dashboard surfaces the review queue alongside new content — mastery isn't a one-time event.
Local-first, offline-capable
A ~10 MB Tauri 2 binary. BKT, SM-2, SQLite (WAL), and vector search all run in-process. No telemetry, no internet required for the core loop.
Bring your own AI key
Claude, ChatGPT, or Gemini — paste a key or sign in with an existing subscription — or point SkillCoco at a local Ollama instance. No hosted backend to pay for or trust.
The only OSS platform like it
MIT licensed, source-auditable, and the only open-source adaptive learning platform that ships BKT, SM-2, AI integration, and an embedded terminal in one installable binary.
Terminal labs you can use — and build
Hands-on labs run in an isolated Docker sandbox when available, with an automatic host-shell fallback. Every step is checked deterministically first (regex, exit code, file state); an AI judge only steps in for open-ended work.
- Author your own lab-based courses — the pack format is open.
- Ships with bundled starter packs to learn from and remix.
- An embedded PTY terminal (xterm.js), not an iframe hack.
$ kubectl apply -f pod.yaml
pod/nginx-demo created
$ kubectl get pods -o wide
NAME READY STATUS RESTARTS AGE
nginx-demo 1/1 Running 0 4s
✓ Step 3 — Pod is Running
check: commandRegex matched "Running"
→ Step 4 — Inspect the pod's node assignmentOpen core, honestly
Everything on this page is MIT licensed and lives in this one public repository. Commercial products build on top of it — they don't gate what's here.
SkillCoco
The adaptive engine, pack format, lessons, gamification, AI tutor, and terminal labs. Public, MIT licensed, always free.
SkillCoco Pro
The integrated learning environment: embedded terminal + IDE, validated/graded labs, interactive simulators, exam mode.
SkillCoco Hub
Course licensing, verifiable certificates, progress sync, and cohort reporting for educators and teams.
SkillCoco Studio
Course authoring and AI enrichment for educators — transcripts and outlines in, a finished pack out.
Pro, Hub, and Studio previews are coming soon. Get in touch if you want early access.
Get SkillCoco
Two ways to run it — download the signed build, or build from source.
Download the app
macOS (Apple Silicon) · signed & notarized · latest release
Download .dmgSee all releases for other platforms and changelogs.
Build from source
git clone https://github.com/skillcoco/skillcoco.git
cd skillcoco
pnpm install
cargo build --manifest-path src-tauri/Cargo.toml
pnpm tauri devRequires Node 18+, pnpm 8+, and Rust stable.
Growing in the open
Every commit, issue, and decision is public. If SkillCoco is useful to you, a star helps others find it.
Ready to build your first course?
Free, offline-capable, and yours to audit line by line.