// Use when the user wants to sit down and learn, continue a lesson, practice a skill, or work through a learning project. Triggers on: "let's learn", "continue my lesson", "start learning", "practice", "I have time to learn", or "pick up where I left off".
Use when the user is done learning for the day, wants to save progress, or wants to wrap up a lesson. Also triggered by /learn when postponing or abandoning. Updates the learning intelligence layer (insights.md) after each session. Triggers on: "end session", "done learning", "save progress", "wrap up", "end-learn", "I'm done for today".
Internal-only skill — never called directly by users. Provides the Learning Intelligence Layer. Two modes: "build context" (called by /learn at session start) reads learning history and produces a learner context; "update insights" (called by /end-learn after session close) updates the persistent insights.md summary.
Use when the user wants to review what they've learned, see progress on their gaps, get a summary of recent learning sessions, or wants article ideas based on learnings. Triggers on: "learning report", "what have I learned", "progress report", "summarize my learning", or "learning summary for the last month".
Use when the user wants a fast status check on their learning without a full report. Triggers on: "learning status", "where am I", "what's my learning state", "how's my learning going", or "check my progress".
Use when the user wants to find what to learn next, explore learning topics, get project ideas for a skill gap, or add a learning idea to their backlog. Triggers on: "what should I learn", "suggest learning projects", "I want to learn X", "add to my learning backlog", or "find me resources on X".
Use at the end of a Claude Code session to capture project-specific friction as proposed CLAUDE.md edits. Triggered by /debrief (no args) to scan the full session, or /debrief <hint> to focus on a specific area (e.g., /debrief the API naming convention thing). Returning "session looks clean" with no proposals is a first-class outcome, not a fallback. Triggers on: "/debrief", "debrief the session", "wrap up this session", "what did we learn", "any rules to capture", "should we update CLAUDE.md".
| name | learn |
| description | Use when the user wants to sit down and learn, continue a lesson, practice a skill, or work through a learning project. Triggers on: "let's learn", "continue my lesson", "start learning", "practice", "I have time to learn", or "pick up where I left off". |
| license | MIT |
| version | 0.1.0 |
| metadata | {"author":"david","tags":"learning, session, lesson, practice, teaching, milestones, hands-on, reading"} |
The core learning experience. Start a new lesson from the backlog, resume the current one, or pick up a postponed lesson. Deliver content adapted to the user's chosen style, track observations silently, and pace to the user's session length.
Profile must exist at ~/.claude/learning/profile.md. If not, tell the user:
"No learning profile found. Run /init-learning first to set up your profile."
Read these files:
~/.claude/learning/profile.md — style preference, session_length~/.claude/learning/lessons/current.md — active lesson (may not exist)~/.claude/learning/backlog.md — available items~/.claude/learning/lessons/archive/ for lessons with status: postponedIf current lesson exists (current.md is present):
/end-learn closing flow (ask reason if abandoning), then proceed to "no current lesson" flow belowIf no current lesson:
archive/ back to lessons/current.md, resume from last milestonelessons/current.mdInvoke the learning-insights skill in "build context" mode. It reads archived lessons,
session logs, and insights.md to produce a learner context.
If fuzzy_topics found, flag them to the user: "Before we start — these came up as fuzzy in past sessions:
Wait for the user's choice. If they want to review, spend a few minutes on the topic (a quick question or explanation), then proceed to the lesson. If they want to dive in, proceed.
If cross_topic_patterns found, mention briefly: "Heads up: I've noticed [theme] keeps coming up across different topics. I'll keep an eye on that today."
This is informational only — do not wait for a response, proceed to Session Start.
Difficulty adjustment feeds into milestone generation in step 4:
Ask: "Hands-on or reading today?" (show the default from profile, user can override)
If hands-on, also ask: "Where should we create the project?" (suggest a path, user confirms)
project_path in current.mdAdapt to the chosen style for this session:
Challenge-based:
Example challenge prompt:
**Milestone 2: Fetch real API data**
Your task: Replace the mock data with real API calls.
**Success criteria:**
- App fetches data from the /api/metrics endpoint on mount
- Shows a loading spinner while fetching
- Displays an error message if the request fails
- Table renders with real data on success
**Hints available** — just ask if you get stuck. Try it yourself first!
Guided:
Reading:
Throughout the session, silently track these observations (do NOT show them to the user during the session — they are for /end-learn):
Keep these observations in your conversation context as a mental running list. They will be synthesized by /end-learn. Do NOT write them to any file during the session.
Example internal observation (never shown to user):
Observation: Hint given — user confused useState return value, explained it's [value, setter].
Observation: Nailed it — destructuring props on first try without help.
Observation: Struggle — tried to mutate state directly 3 times before understanding immutability.
session_length: "You're at about ~Xm — good stopping point after this exercise, or keep going?"When the user wants to stop (or you suggest stopping and they agree):
/end-learn to save your progress and capture what you learned."/end-learn handles all persistenceWhen creating a new lesson from a backlog item:
session_length)current.md/new-learningcurrent.md session log between sessions — append, don't overwrite