# Whiteboard Help > Show the quick-reference guide for the whiteboard planning system (whiteboard, hot-seat, napkin, connotation-cop) — the flow, what to say, and the rules. Use when the user asks "whiteboard help", "how do I use the whiteboard", "what's the flow", "remind me how whiteboard works", "which skill do I use", or seems unsure how to start or continue a whiteboard effort. This skill only DISPLAYS the guide — it never starts a board, resolves tickets, or does any work. - Skill: `theycallmeholla/whiteboard-help` (Agent Skill) - Install (CLI): `npx skillmds add theycallmeholla/whiteboard-help` - Raw SKILL.md: https://api.skillmd.com/api/skills/theycallmeholla/whiteboard-help/raw - Safety review: pending (external: skill-scanner PASS, skillspector PASS) - Works with: Claude Code, Claude.ai, OpenAI Codex - Category: Coding & Dev Tools - Author: Theycallmeholla (https://skillmd.com/u/theycallmeholla) - Updated: 2026-08-19 - Page: https://skillmd.com/skills/theycallmeholla/whiteboard-help --- # Whiteboard Help Display the guide below to the user, adapted to what they asked. If they asked a specific question ("how do I close a board?"), answer it from the guide and show just the relevant part. If they asked generally, show the whole thing. **Do not start any work** — this skill is a cheat-sheet, nothing else. If they want to actually begin, tell them what to say and let them say it. --- ## The one-liner Big fuzzy idea → put it on the **whiteboard** → work the tickets, stopping only when it needs *you* → **snapshot** the board into a spec when nothing's left to decide. ## The flow **1. Draw** — one session, run once per effort Say: **"whiteboard this: [your idea]"** (in a repo with GitHub Issues) What happens: Claude hot-seats you to pin the destination, creates the board issue + tickets, parks the blurry stuff in the parking lot. Then stops. **2. Work** — repeat until nothing's left to decide Say: **"work the board"** — or **"run the board"** when you're walking away, or "work one ticket" for exactly one. What happens: Claude claims the next open ticket and resolves it. Ticket type decides the tool, and whether it needs you: - hot-seat ticket → Claude questions you, one at a time, with its pick attached — **needs you** - napkin ticket → Claude builds a throwaway sketch; the verdict is yours — **build alone, judge with you** - research ticket → Claude reads docs alone, links a summary — **doesn't need you** - task ticket → a prep chore that unblocks a decision; the ticket says `Runs: agent` or `Runs: human` Then it updates the board and checks what's next. **Next ticket needs nobody → it just keeps going.** Next ticket needs you → it stops and tells you what it wants. You should never come back to a session idling on a ticket it could have done itself. **3. Snapshot** — one session, run once at the end Say: **"snapshot the board"** What happens: Claude verifies nothing's left (no open tickets, empty parking lot), writes the spec from all the decisions, links it, closes the board. Done — hand the spec to whoever builds. ## Solo use (no board needed) - **"hot seat me on [decision]"** — stress-test any plan, one question at a time - **"napkin this: [design question]"** — throwaway code that answers one question - **"connotation-cop: [term fight]"** — settle what a word means, book it in the glossary ## The rules that matter - **The sticky test:** can you write the question on one sticky note, precisely? Yes → it's a ticket. Still a paragraph of maybes → parking lot. - **It stops for you, not for tickets.** A session ends when you're the blocker — never just because a ticket closed. Between every ticket it re-reads the board, so the picture stays current. - **The board plans, it doesn't build.** Open decision left? That's the next ticket. None left? Snapshot. - **Claim before work.** Assignee on the ticket = claimed. Unassigned + unblocked = up for grabs. - **Decisions live on tickets.** The board only gists and links — never holds detail. - **Every *stop* ends with a sign-off:** board count, what got done, **why it stopped**, what's waiting on you, and what to say. You should never have to ask "what's next" — if a session stops without the sign-off, say "sign off" and you'll get it. ## Where things live - The board: a GitHub issue labelled `whiteboard:map`; tickets are its sub-issues - Decisions: resolution comments on closed tickets, indexed on the board - Big permanent decisions: ADRs in `docs/adr/` (connotation-cop writes them) - Vocabulary: `CONTEXT.md` at the repo root - The final spec: wherever the board's Destination says ## Stuck? - "Which ticket next?" → say "work the board" — Claude picks the first unblocked, unclaimed one - "It stopped and I don't know why" → the sign-off's **Stopped because** line says: you're the blocker, the board is fanning out faster than it's closing, a ticket drifted off the destination, or context ran thin. Only the first one needs an answer from you — the rest just need "run the board" again - "It stopped on something it could have done itself" → that's a bug now, not the design. The rule is: needs you → stop, doesn't → keep going - "Ticket feels huge" → it was mis-sized; Claude splits it into sharp sub-tickets - "Someone abandoned a ticket" → claimed + 3 days silent = presumed dead; Claude asks you before unclaiming - "New idea mid-effort but it's past the destination" → it goes to Out of scope, not the board