Learn to think before you prompt.
Demanding, Socratic, AI-era courses and tools that build genuine competence — the tutor refuses to do the work for you.
Website · Courses · Philosophy
We build hands-on courses and tools fronted by a demanding Socratic tutor that runs inside your AI coding agent. It won't write your code, hand you answers, or confirm a flag. It enforces the discipline that turns months of work into a body of evidence you can defend in an interview.
| Project | What it is | For | Status |
|---|---|---|---|
| Vigil | 12-month, lab-driven cybersecurity course, novice → practitioner | Aspiring security pros | Active |
| Agentwright | 12-month agentic-engineering course, zero coding → deployed AI systems | Future AI engineers | Active |
| Muster | Pattern-first LeetCode & coding-interview mastery in Python | Interview prep | Active |
| Wachturm | Self-hosted Tier-1 SOC analyst training (Wazuh, real tools) | Blue-team learners | Active |
| Lernen | CLI tutor for true fluency in one programming language | Vibe coders going deep | Pre-1.0 |
| Lernen Praxis | In-host skill for safe, effective agentic-coding-tool use | Agent operators | In design |
- Concept first, command second — no tool runs until you can explain what it does to the wire, the disk, or the kernel.
- The struggle floor — a fixed window of unaided work before any hint or walkthrough.
- Hints, not answers — a rung-by-rung hint ladder; the tutor never hands you the solution or confirms a flag.
- Spaced repetition is the curriculum — prior work pulled back and re-solved cold, on a schedule.
- AI is an apprentice's tool, not a crutch — fundamentals stay AI-free; augmentation unlocks under provenance discipline.
Curriculum content is generally CC BY (some BY-SA); code is MIT/AGPL per repo — see each repository for specifics. Contributions are welcome; forking is the recommended path for almost everyone.