weorbitant/orbitant-poc-skills · Archived

end-learn

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".

First seen Apr 22, 2026

Installation

$ npx skills add weorbitant/orbitant-poc-skills --skill end-learn

Stronger alternatives

This repository is archived — consider an actively maintained alternative.

Similar popular skills

Related neighbors and high-traction skills in the same topics — useful to compare before installing.

Also in this package

Other skills from weorbitant/orbitant-poc-skills.

npx skills add weorbitant/orbitant-poc-skills

Browse all from weorbitant/orbitant-poc-skills

More details

Agent compatibility

Declared targets from SKILL.md / docs. Unmarked agents are not listed — the skill may still install via the CLI.

Claude Code Declared
Cursor Not declared
Codex Not declared
GitHub Copilot Not declared
Windsurf Not declared
Gemini CLI Not declared
Cline Not declared
OpenCode Not declared

Repository health

License MIT
Default branch main
Open issues 1
Status Archived

Skill metadata

Parsed from SKILL.md frontmatter.

Version0.1.0
LicenseMIT
Declared agents claude-code
More metadata
author
david
tags
learning, session, log, debrief, observations, progress, wrap-up

Package contents

Files included with this skill beyond the listing page.

  • skill md SKILL.md 5,491 B
  • docs SUMMARY.md 356 B

History

  1. First seen on skills.sh
  2. First recorded snapshot · 4 installs

SKILL.md

Overview

Close a learning session by synthesizing Claude's silent observations into a debrief the user reviews. Write the session log, update lesson state, and archive completed lessons.

When to Use

  • User finishes a learning session
  • User wants to save progress and stop
  • Called by /learn when postponing or abandoning a lesson

Prerequisites

There must be an active lesson at ~/.claude/learning/lessons/current.md. If not, tell the user: "No active lesson found. Nothing to close."

Process

1. Read State

  • ~/.claude/learning/lessons/current.md — current lesson state
  • Session observations from the current conversation (hints given, struggles, successes)

2. Present Draft Debrief

Synthesize observations into a structured draft. Use this exact format:

Here's what I noticed this session:

**What you worked on**
- [summary of milestone(s) attempted and outcomes]

**Key learnings**
- [things the user demonstrated understanding of]
- [concepts that "clicked" based on observed aha moments]

**Struggled with**
- [areas where hints were needed]
- [repeated mistakes or confusion points]

**Still fuzzy on**
- [unresolved questions]
- [topics where understanding seemed incomplete]

If a section is empty (e.g., no struggles), include it with "Nothing notable — solid session!" rather than omitting it.

Ask: "Edit anything, add something I missed, or looks good?"

Wait for the user to confirm or make edits. Do NOT proceed until the user responds.

3. Determine Lesson Status

If all milestones are completed:

  • Congratulate the user
  • Status: completed

If not all milestones are done:

  • Ask: "What should we do with this lesson?"

- Continue next time — keep as current.md, no archive - Postpone — archive with status: postponed - Abandon — ask for a brief reason, archive with status: abandoned

4. Write Session Log

Write to ~/.claude/learning/logs/YYYY-MM-DD.md:

  • If the file already exists (same-day session), use suffix: YYYY-MM-DD-2.md, YYYY-MM-DD-3.md, etc.
  • Use the session log format from ~/.claude/skills/learning-data-spec.md
  • Content comes from the confirmed debrief

5. Update Lesson State

If continuing next time:

  • Update milestones_completed in current.md frontmatter
  • Mark completed milestones as [x] in the milestones list
  • Append a session log entry to current.md:

`` ### Session N — YYYY-MM-DD - Completed: [milestones] - Struggled with: [summary] - Next: [what to tackle next] ``

If completed:

  • Move current.md to lessons/archive/YYYY-MM-DD-<topic-slug>.md
  • Set frontmatter: status: completed, closed: YYYY-MM-DD, sessions: N
  • Add ## Key Learnings and ## Still Fuzzy On sections from accumulated observations
  • Mark the corresponding backlog item as [x] done in backlog.md
  • Delete current.md

If postponed:

  • Move current.md to lessons/archive/YYYY-MM-DD-<topic-slug>.md
  • Set frontmatter: status: postponed, closed: YYYY-MM-DD
  • Preserve all milestone progress and session logs
  • Delete current.md

If abandoned:

  • Move current.md to lessons/archive/YYYY-MM-DD-<topic-slug>.md
  • Set frontmatter: status: abandoned, closed: YYYY-MM-DD, abandoned_reason: <reason>
  • Preserve all milestone progress and session logs
  • Delete current.md

5.5. Update Insights

Invoke the learning-insights skill in "update insights" mode.

Pass the session's observations from the confirmed debrief:

  • Struggles (from "Struggled with" section)
  • Key learnings (from "Key learnings" section)
  • Still fuzzy topics (from "Still fuzzy on" section)
  • Hints given during the session (from silent observation tracking)
  • Any incidental code observations noted during the session

The learning-insights skill will:

  1. Update ~/.claude/learning/insights.md (create it if first time)
  2. Return a one-liner if something meaningful changed

If a one-liner is returned, show it to the user before the confirmation message in step 6. Examples:

  • "Insight updated: LCEL chain composition moved to confirmed strength"
  • "Pattern detected: state management has come up in 3 sessions now"

If nothing meaningful changed, say nothing — proceed to step 6 silently.

6. Confirm

Tell the user what was saved:

  • "Session log saved to logs/YYYY-MM-DD.md"
  • If archived: "Lesson archived as [status] to lessons/archive/[filename]"
  • If continuing: "Progress saved. Run /learn next time to continue from milestone X."

Important

  • Always present the debrief and get user confirmation BEFORE writing any files
  • Observations come from the current conversation — if /end-learn is called in a new

conversation (without a preceding /learn), ask the user to describe what they did

  • Create directories if they don't exist (logs/, lessons/archive/)
  • The slug for archive filenames: lowercase, hyphens, max 50 chars (e.g., "react-dashboard-api-metrics")