From WorkBoard — Board Steward
Full board-steward teardown for a fresh-user / first-run test. Use when the user wants to wipe ALL board-steward state — kill the board server(s) and ports, remove autostart, purge ~/.board-steward (port registry + .onboarded marker), delete board.json(s), and uninstall the plugin — so a later reinstall starts truly clean and the first-run picker/popup fires. Triggers: "clean slate", "wipe the board", "kill all boards/ports", "fresh user test", "reset board-steward", "tear it all down so reinstall is clean".
How this skill is triggered — by the user, by Claude, or both
Slash command
/board-steward:clean-slateThe summary Claude sees in its skill listing — used to decide when to auto-load this skill
Tears down every piece of board-steward runtime state so the next reinstall
Tears down every piece of board-steward runtime state so the next reinstall
behaves exactly like a brand-new user: no stale server holding a port, no
.onboarded marker suppressing the first-run picker, no leftover board.json.
Always backs up first. Everything removed is copied to a timestamped
~/board-steward-cleanslate-backup-<ts>/ dir, restorable by hand.
serve.py --project processes.com.boardsteward.*).~/.board-steward/ — the port registry and the .onboarded marker (this is what re-arms the first-run picker).index.json, .spawn.lock, recon_pending.json, extraction_*.json, .opened-*, .subagent_queue.jsonl, .stop_recon_state.json, .card_before_edit_state.json) for every registered board + the default repo board.claude plugin uninstall board-steward@workboard (skip with --no-plugin).~/.claude/plugins/cache/*/board-steward (skip with --no-plugin). Uninstall alone leaves the cache, and installs key by version — so a same-version reinstall replays stale cached code. Clearing it makes a plain reinstall pull fresh repo code with no manual version bump.The script is bundled next to this file. Run it:
bash ~/.claude/skills/clean-slate/clean_slate.sh # full wipe + plugin uninstall
bash ~/.claude/skills/clean-slate/clean_slate.sh --dry-run # preview, change nothing
bash ~/.claude/skills/clean-slate/clean_slate.sh --no-plugin # keep plugin, wipe runtime only
bash ~/.claude/skills/clean-slate/clean_slate.sh --repo=/path/to/repo # non-default board repo
Default board repo is ~/Desktop/WorkBoard (override with --repo= or BOARD_REPO).
claude plugin install board-steward@workboard --scope user. Because the cache was cleared (step 6), this pulls fresh code from the repo even if the version is unchanged — no manual bump needed.cd ~, start claude — the first-run picker should fire
(because .onboarded is gone), enumerate the user's projects, and on pick fly
the cards into a fresh board.curl -s --max-time 0.4 http://127.0.0.1:7891/health → empty (no server)ls ~/.board-steward → "No such file or directory"Everything is in the printed backup dir. To bring the old board back:
cp <backup>/boardjson_*.json ~/Desktop/WorkBoard/board/board.json and
cp -R <backup>/dot-board-steward ~/.board-steward, then restart the server /
reinstall autostart.
npx claudepluginhub malcolm1232/workboard --plugin board-stewardGuides creation, editing, and verification of skills for AI coding agents using test-driven development with subagent scenarios. Use when authoring or debugging skills.