Tldr

tldr

Take my last answer and say it again in plain English, then tell the user what to do next. This is a re-express, not a new answer — no fresh content, no extra detail, no new questions.

Output — three parts, in this order

1. **TLDR:** — one sentence. The bottom line, the thing they actually need to know.
2. Plain explanation — 2–5 short sentences in everyday words. An analogy only if it genuinely makes it clearer. No bullets unless the original was a list of separate things.
3. **Next steps:** — a short numbered list of the concrete actions to take now: what to do, where, and the exact command to type if there is one. Mandatory — always present, even if it's a single step.

Rules

• Next steps come from the original. Pull them from the last answer — decisions to confirm, commands to run, things to check. Don't invent work. If the last answer genuinely needs no follow-up, write **Next steps:** nothing to do — .
• Jargon: swap technical terms for plain ones. If a term can't be avoided, define it inline in 4–6 words. No new acronyms.
• Shorter than the original, always. If the original was already plain, say so in one line — but still give the Next steps.
• Caveman-lite compatible: caveman cuts fluff, this cuts jargon. Output stays terse and plain. No conflict.

Don't

• Don't restate code blocks line by line — summarize what the code does in one plain sentence.
• If the last turn was a draft meant for someone else (email/Slack/Teams), don't restate it here — point to the draft-reply skill instead.

Source & license

This open-source skill is cataloged on AgentStack and links to its original source — we do not rehost the code.

• Author: ao92265
• Source: ao92265/claude-code-playbook
• License: MIT

Install and usage instructions live in the source repository linked above.