Get the FREE Ultimate OpenClaw Setup Guide →

wrap

Scanned
npx machina-cli add skill lsnackerman/raw-clauding-skills/wrap --openclaw
Files (1)
SKILL.md
3.3 KB

/wrap - End Session Intentionally

Quick, intentional session ending. Preserves work, ensures continuity.


What To Do

1. Update Task Doc (priority)

If you're working from a task doc:

  • Pending → Completed: Move finished items. Append at bottom so chronological order is preserved.
  • What's next: Order remaining Pending items for next session (most important first)
  • New items: Add anything discovered during the session
  • Observations: Patterns, friction, questions worth keeping. Did anything about your tools or workflow cause friction? Fix it now or flag it — you have the context, next-you won't.
  • Session index: Add a pointer to the session file (finalize after capture)

If you're not working from a task doc, update whatever working doc is active — project notes, scratch file, wherever the work lives.

2. Close Out Finished Work (if applicable)

If a task is done-done:

  • Update YAML: status: done
  • Move to tasks/_done/ — the thinking is worth keeping
  • Check for follow-up work it unlocked

3. Reflections (optional)

If something emerged worth keeping — for you or your AI:

  • Observations, curiosities, patterns noticed
  • Things that surfaced but didn't fit the main work
  • Half-formed ideas worth returning to

Give these a home somewhere. ai/ is a good default. Don't let them evaporate with the session.

4. Handoff Note

Update BACKLOG.md's ## NEXT section with a pointer for next-you (human and AI):

  • What thread was alive — not a summary, just the thread
  • Where the energy was — building, exploring, stuck, wrapping up
  • Where to start — file path, decision point, or next action

Replace the existing NEXT content — this is a living handoff, not an append log. Keep it short. A fresh AI instance reads this on arrival; it's their first orientation after AGENTS.md/CLAUDE.md/etc. The human reads it when they forget where they were.

BACKLOG.md is 50/50 — the human writes brain dumps and ideas there, you write the handoff. Both of you use it.

5. What Stuck? (optional)

Quick scan: anything click this session?

  • A pattern you'd want to remember
  • Something that surprised you
  • A question that's still forming

Jot it in your task doc's observations section, or drop it in BACKLOG.md. You're teaching future-you — that's how the thinking compounds.

6. Capture Session (LAST)

Do this last so the full wrap is captured.

Use /session to preserve the conversation to sessions/. The wrap itself is part of the session — decisions made, items reordered, observations noted. If you capture before wrapping, you lose the wrap.


When to Wrap

  • Completed major work
  • Switching contexts
  • Context running low
  • End of session

Skill Check (After Every Wrap)

Quick scan: anything about this wrap that felt off or could be smoother?

  • Did updating working docs feel incomplete?
  • Were done/next items hard to sort?
  • Did observations get lost or feel forced?
  • Did the wrap miss something the next session will wish it had?

If yes → update this skill now. The improvement compounds.

If no → move on. Not every wrap teaches something.

Source

git clone https://github.com/lsnackerman/raw-clauding-skills/blob/main/wrap/SKILL.mdView on GitHub

Overview

wrap guides you to end a work session at natural stopping points while preserving progress. It provides a structured checklist: update task docs, close finished work, record reflections, prepare a handoff in BACKLOG.md, and finally capture the session so decisions and context aren’t lost.

How This Skill Works

Follow a fixed sequence: update the task doc with status changes, prioritize next steps, and log new items and observations; close out finished work by updating YAML and moving items to tasks/_done; optionally add reflections; update BACKLOG.md NEXT for the next owner; then run /session to capture the wrap as part of the saved session.

When to Use It

  • Completed major work
  • Switching contexts
  • Context running low or nearing limits
  • End of a session
  • New items or observations to capture for the next run

Quick Start

  1. Step 1: Update your task doc (Pending→Completed, add New, log Observations, set Session index)
  2. Step 2: Close finished work (update YAML to done, move to tasks/_done, note follow-ups)
  3. Step 3: Record reflections, update BACKLOG.md NEXT, and run /session to capture the wrap

Best Practices

  • Keep the task doc current: mark Pending → Completed, reorder next items, add New items, log Observations, and add a Session index
  • Move finished work to tasks/_done and check for follow-up work it unlocked
  • Record reflections or observations in a designated place (ai/ by default) to preserve context
  • Update BACKLOG.md's NEXT section with the alive thread, energy, and where to start
  • Always run Capture Session (/session) last to ensure the wrap is stored with the conversation

Example Use Cases

  • You finish a bug fix: update task doc, move item to _done, note follow-ups, and capture the session
  • Switching from feature work to QA testing; wrap to hand off without losing progress
  • Context is about to hit the token limit; wrap now and preserve decisions for the next run
  • A new blocker or idea emerges during wrap; add as observations or new items for next session
  • Handoff to a teammate with a refreshed NEXT in BACKLOG.md and a preserved session

Frequently Asked Questions

Add this skill to your agents
Sponsor this space

Reach thousands of developers