← agent skills / learning-mode

learning-mode

Turn a Claude Code session into a learn-by-doing coding tutorial with a spaced-repetition review log.

github.com/Osipchuk/agent-skills/skills/learning-mode ↗ SKILL.md
learning

Install

install · cli 
$ curl -fsSL https://raw.githubusercontent.com/Osipchuk/agent-skills/main/install.sh | sh -s -- learning-mode

One line, works with any agent — installs just this skill into ~/.claude/skills/. Needs only curl; it fetches uv and the askill CLI if they're missing.

Installs skills/learning-mode/SKILL.md into ~/.claude/skills/learning-mode/; Claude reads it on the next session start. The plugin tab pulls the whole library instead.

When it fires

Fires when you ask to learn, practice, or be coached on something during a coding session — and at the start of any session in a repo that already has a .claude/learning/ plan with topics due for review.

# skills/learning-mode/SKILL.md — frontmatter
name: learning-mode
description: Turn a regular Claude Code session into a learn-by-doing coding tutorial. The user keeps shipping real work, but at deliberate moments Claude hands off a focused chunk for the user to write by hand, then reviews it. Maintains a per-repo learning plan and a spaced-repetition log under .claude/learning/ so topics resurface at the right time. Use this whenever the user mentions wanting to learn, practice, train, "do it themselves", improve at programming, or asks Claude to coach, teach, tutor, mentor, or pair-learn in a coding context. Also consult it at the start of any session in a repo containing .claude/learning/ — there may be topics due for review or homework in flight. Trigger phrases include "coach me", "teach me X", "I want to learn Y", "help me practice Z", "let me try it myself", "review what I wrote", "give me a task", "what should I learn next", and any signal the user wants to grow as a developer rather than just ship code. Do not use when the user only wants the work done for them.

What it does

  • — Hands you one atomic piece to write yourself, then reviews it like a senior pair-programmer.
  • — Splits the work instead of stalling — Claude ships the rest of the feature while your piece is open, and never touches your file.
  • — Keeps a per-repo learning plan and a spaced-repetition log under .claude/learning/.
  • — Resurfaces due topics at the right time, so what you practiced sticks before you forget it.

Example

example · coaching handoff in a real repo 2 turns · 90 tok
input
user
Teach me Python decorators while we build this retry helper.
↓ skill fires
skill
[learning-mode] handoff — write `retry` yourself; brief in .claude/learning/active-task.md
output
assistant
Scaffold's in place with `timed` next to it as the worked example. Your turn: write `retry`. Meanwhile I wired the call sites and left you a red test — your file's untouched. Ping me when you want a review.
Without learning-mode, Claude would just write the decorator for you — no practice, no retention.