name

boardroom-creator

description

Scaffold a new AI boardroom — interviews the user (AskUserQuestion) about the board's domain, seats, and context, then generates a new board skill from the proven startup-board template plus the company/domain context folder. Use when the user says "create a boardroom", "new board", "/boardroom-creator", or wants a board for a new company or a new domain (product board, content board, personal board).

Boardroom Creator

Create a new boardroom: either (a) a new COMPANY context folder for an existing board skill, or (b) a new KIND of board (new domain, new slate) as its own skill. Collect everything through batched AskUserQuestion interviews, then scaffold files.

The master persona library lives in this skill's personas/ directory. Generated boards get COPIES of the personas they seat, so every board stays self-contained.

Step 1 — Interview: what kind of boardroom?

Ask (AskUserQuestion, batched — up to 4 questions per call, not one at a time):

New company or new board type?
- "Company for an existing board" → go to Step 2A.
- "New kind of board" (product board, content board, personal board, ...) → Step 2B.
Name/slug — company slug (e.g. regalscene) or board name (e.g. content-board).
Where context lives — default ./boardrooms/<slug>/ under the current directory; offer $BOARDROOM_HOME if set.

Step 2A — New company for an existing board

Interview (batched):

What the company does, stage, market, main competitors (open text).
Where the numbers are (revenue, pricing, pipeline) — file to import, or dictate now.
Founder values and hard boundaries (things the board must treat as constraints).
Slate: keep the board's default 8 seats, or swap? For each swap, offer personas from the master library; for a person not in the library, write a new ~40-line profile (worldview, decision heuristics, known frameworks, biases, communication style, always-pushes-back-on, what-makes-them-vote-NO) and save it to the master library first, then copy it into the board.

Scaffold boardrooms/<slug>/:

business-context.md, numbers.md, values-and-boundaries.md — pre-filled from the interview; anything unknown gets an explicit TODO(user): marker.
board.md only if the slate differs from the board skill's default.
meetings/ (empty).

Then tell the user which TODOs must be filled before the first meeting.

Step 2B — New kind of board (new skill)

Interview (batched):

Domain and decisions it rules on (what questions come to this board?).
The seats: propose a slate of 6-8 with mandates fitted to the domain — pull from the master library where personas fit, propose new people where they don't. Let the user swap per seat. Write new persona profiles into the master library, then copy.
What context files this domain needs (the startup default is business-context / numbers / values-and-boundaries; a content board might need audience.md and voice.md instead). Define 2-4 required files and what goes in each.
Values/boundaries that apply across all meetings of this board.

Scaffold ~/.claude/skills/<board-name>/:

SKILL.md — copy the startup-board SKILL.md structure verbatim (it is the proven template: Step 0 resolve → Round 1 independent → Round 2 rebuttals → synthesis → dashboard-on-request → rules), replacing: the frontmatter name/description, the required context files list, and the slate. Keep the two-round mechanics, the vote block formats, the ABSTAIN handling, and the never-invent-numbers rule unchanged.
board.md — the new slate.
personas/ — copies of the seated personas from the master library.

Then create the first context folder for it (Step 2A flow) if the user wants one now.

Rules

Batch interview questions with AskUserQuestion (multiple questions per call); only fall back to open text where options don't fit.
Never invent business facts to fill context files — unknowns become TODO(user):.
New personas always land in the master library first, then get copied out - the library only grows.
Do not modify an existing board skill when adding a company - companies are context folders, not skill edits.

name

boardroom-creator

description

Boardroom Creator

The master persona library lives in this skill's personas/ directory. Generated boards get COPIES of the personas they seat, so every board stays self-contained.

Step 1 — Interview: what kind of boardroom?

Ask (AskUserQuestion, batched — up to 4 questions per call, not one at a time):

New company or new board type?
- "Company for an existing board" → go to Step 2A.
- "New kind of board" (product board, content board, personal board, ...) → Step 2B.
Name/slug — company slug (e.g. regalscene) or board name (e.g. content-board).
Where context lives — default ./boardrooms/<slug>/ under the current directory; offer $BOARDROOM_HOME if set.

Step 2A — New company for an existing board

Interview (batched):

What the company does, stage, market, main competitors (open text).
Where the numbers are (revenue, pricing, pipeline) — file to import, or dictate now.
Founder values and hard boundaries (things the board must treat as constraints).
Slate: keep the board's default 8 seats, or swap? For each swap, offer personas from the master library; for a person not in the library, write a new ~40-line profile (worldview, decision heuristics, known frameworks, biases, communication style, always-pushes-back-on, what-makes-them-vote-NO) and save it to the master library first, then copy it into the board.

Scaffold boardrooms/<slug>/:

business-context.md, numbers.md, values-and-boundaries.md — pre-filled from the interview; anything unknown gets an explicit TODO(user): marker.
board.md only if the slate differs from the board skill's default.
meetings/ (empty).

Then tell the user which TODOs must be filled before the first meeting.

Step 2B — New kind of board (new skill)

Interview (batched):

Domain and decisions it rules on (what questions come to this board?).
The seats: propose a slate of 6-8 with mandates fitted to the domain — pull from the master library where personas fit, propose new people where they don't. Let the user swap per seat. Write new persona profiles into the master library, then copy.
What context files this domain needs (the startup default is business-context / numbers / values-and-boundaries; a content board might need audience.md and voice.md instead). Define 2-4 required files and what goes in each.
Values/boundaries that apply across all meetings of this board.

Scaffold ~/.claude/skills/<board-name>/:

SKILL.md — copy the startup-board SKILL.md structure verbatim (it is the proven template: Step 0 resolve → Round 1 independent → Round 2 rebuttals → synthesis → dashboard-on-request → rules), replacing: the frontmatter name/description, the required context files list, and the slate. Keep the two-round mechanics, the vote block formats, the ABSTAIN handling, and the never-invent-numbers rule unchanged.
board.md — the new slate.
personas/ — copies of the seated personas from the master library.

Then create the first context folder for it (Step 2A flow) if the user wants one now.

Rules

Batch interview questions with AskUserQuestion (multiple questions per call); only fall back to open text where options don't fit.
Never invent business facts to fill context files — unknowns become TODO(user):.
New personas always land in the master library first, then get copied out - the library only grows.
Do not modify an existing board skill when adding a company - companies are context folders, not skill edits.

boardroom-creator

Boardroom Creator

Step 1 — Interview: what kind of boardroom?

Step 2A — New company for an existing board

Step 2B — New kind of board (new skill)

Rules

同仓库更多 Skills

同仓库更多 Skills

Boardroom Creator

Step 1 — Interview: what kind of boardroom?

Step 2A — New company for an existing board

Step 2B — New kind of board (new skill)

Rules