Browse / Teaching & study / mentor-os
mentor-os
Tammo-Feldmann
A Claude Code workspace that acts as a persistent AI mentor — it learns who you are through editable context files and runs structured practice sessions calibrated to your goals, gaps, and learning style.
Open Tammo-Feldmann/mentor-os → Checked against the four rules.
Built for one agent
ships CLAUDE.md and nothing for other agents.
In its own words
A personal AI mentoring system for serious technical skill development.
The problem with grinding
Grinding through tutorials and LeetCode problems is efficient for content delivery but not always ideal for skill building. What actually works is what good human mentors do: they get you. They remember what clicked last time, what you're avoiding, what level of challenge is productive vs. demoralizing, and what the right next thing to push on is.
This system gives an AI agent that same persistent context — but with one key difference: you own the context files. You can read them, edit them, and redirect the agent whenever the direction feels off.
How it works
The macro direction is yours. The session details are the agent's.
You decide what matters and where to focus. The agent handles picking the right challenge, running the Socratic session, and writing the log afterward. Over time, the logs and pattern files accumulate into a detailed picture of how you learn and what you know — which makes every next session better.
Getting started
1. Clone the repo and open it in Claude Code
The agent will interview you about your background, goals, strengths, gaps, and learning preferences — then generate your profile and goals files.
3. Start your first lesson
The agent picks the best topic based on your context and goals, or you can specify one: lesson: system design.
Quoting 58% of the README. Read the rest on GitHub ↗
Read from the repository's git tree on 2026-08-17.