Skip to content
Public template

About

A personal daily learning system powered by AI agents.

Resources

Stars

5 stars

Watchers

0 watching

Forks

Latest commit

 

History

25 Commits

Folders and files

Repository files navigation

learn-with-ai

A personal daily learning system powered by AI agents. 10–60 minutes a day — short days consolidate, long days extend. Real progress either way.

Works for any topic — software architecture, distributed systems, product management, languages, anything.


Setup

  1. Clone or download this repo
  2. Open CURRICULUM.md — fill in your background, goal, real projects, and what you want to learn
  3. Open this repo in your AI agent (e.g. Claude Code, Cursor, or any agent that reads AGENTS.md)
  4. Start your first session by typing learn

That's it. The agent reads your curriculum and guides everything from there.


How to use

Type one of these in your agent:

learn        → auto-select the next session — asks how much time you have
learn 15     → auto-select with a fixed time budget (also: learn s / m / l)
drill        → ~10 min pure retrieval across due topics (the daily anchor)
quiz         → quiz on the last concept (earliest the day after)
concept      → new concept session  (~20–30 min)
kata         → focused design or coding task  (45+ min, splittable)
deep dive    → trade-off discussion  (45+ min, splittable)
review       → spaced repetition over past concepts  (~20–30 min)

In Claude Code you can also use slash commands: /learn 15, /review — same behavior, guaranteed trigger.

The agent reads PROGRESS.md and CURRICULUM.md automatically. It knows exactly where you are and what comes next.

Every session starts with the time question: S (~10 min) / M (~25 min) / L (45+ min). Ten minutes is a real session — a drill or a quiz — not a skipped day.


Structure

CURRICULUM.md   ← Your personal file: background, projects, curriculum  ← EDIT THIS
AGENTS.md       ← Instructions for the AI agent (generic, no changes needed)
PROGRESS.md     ← Current status, gaps, session log (agent maintains this)

/concepts       ← Concept session notes
/quizzes        ← Quiz results
/katas          ← Design tasks and solutions
/deep-dives     ← Trade-off discussion notes
/review         ← Spaced review logs + drills.md (compact drill log)
/templates      ← Markdown templates for each session type

Tips

  • Short days consolidate, long days extend. New material enters on M/L days; S days strengthen what's there. Both count — a daily 10-minute drill beats a weekly 60-minute marathon for retention.
  • The quiz comes the day after the concept. That's deliberate: it tests what you kept, not what you just heard.
  • Big sessions can be split. If time runs out mid-kata, the agent pauses and resumes next time.
  • Be honest in quizzes. The agent tracks your gaps and revisits them.
  • Use learn each time. The agent picks the right next session automatically.
  • Topics are right-sized automatically. Default per topic: concept + quiz + one application session. The agent expands load-bearing topics to the full cycle and trims small ones — it announces each call, and one word from you overrides it.
  • Reviews happen naturally. After every 4 completed topics, the agent inserts a review.
  • Gap weeks are fine. Come back after a break and the agent picks up exactly where you left off.
  • Real projects make it stick. Add your actual projects to CURRICULUM.md.

Customizing

Everything is driven by CURRICULUM.md. To change your learning path:

  • Edit the phases and topics in CURRICULUM.md

No need to touch AGENTS.md or templates unless you want to change how sessions work.

About

A personal daily learning system powered by AI agents.

Resources

Stars

5 stars

Watchers

0 watching

Forks

Releases

Packages

Contributors