board-meeting
Multi-agent board meeting protocol for strategic decisions. Runs a structured 6-phase deliberation: context loading, independent C-suite contributions (isolated, no cross-pollination), critic analysis, synthesis, founder review, and decision extraction. Use when the user invokes /cs:boardroom, calls a board meeting, or wants structured multi-perspective executive deliberation on a strategic question.
- License 许可
- MIT
- 最近更新
- 2026-08-25
- Stars
- ★ 0
- Forks
- 0
01 前言 Front Matter
- name
- board-meeting
- license
- MIT
- metadata
- { "author": "Alireza Rezvani", "domain": "board-protocol", "updated": 1772668800000, "version": "1.0.0", "category": "c-level", "frameworks": "6-phase-board, two-layer-memory, independent-contributions" }
- description
- Multi-agent board meeting protocol for strategic decisions. Runs a structured 6-phase deliberation: context loading, independent C-suite contributions (isolated, no cross-pollination), critic analysis, synthesis, founder review, and decision extraction. Use when the user invokes /cs:boardroom, calls a board meeting, or wants structured multi-perspective executive deliberation on a strategic question.
03 SKILL.md 正文
Board Meeting Protocol
Structured multi-agent deliberation that prevents groupthink, captures minority views, and produces clean, actionable decisions.
Keywords
board meeting, executive deliberation, strategic decision, C-suite, multi-agent, /cs:boardroom, founder review, decision extraction, independent perspectives
Invoke
/cs:boardroom [topic] — e.g. /cs:boardroom Should we expand to Spain in Q3?
The 6-Phase Protocol
PHASE 1: Context Gathering
- Load
~/.claude/company-context.md - Load Layer 2 approved decisions from
~/.claude/decisions/approved/(Layer 2 ONLY — never raw transcripts) - Reset session state — no bleed from previous conversations
- Present agenda + activated roles → wait for founder confirmation
Chief of Staff selects relevant roles based on topic (not all 14 every time):
| Topic | Activate |
|---|---|
| Market expansion | CEO, CMO, CFO, CRO, COO |
| Product direction | CEO, CPO, CTO, CMO |
| Hiring/org | CEO, CHRO, CFO, COO (+ VPE for eng hiring) |
| Pricing | CMO, CFO, CRO, CPO |
| Technology | CTO, CPO, CFO, CISO |
| Contracts / term sheets / legal exposure | GC, CEO, CFO |
| Data strategy / training-data rights | CDO, CAIO, GC, CISO |
| AI strategy / model selection / AI risk | CAIO, CTO, CDO, CFO |
| Retention / churn / customer success | CCO, CRO, CPO |
| Eng delivery / DORA / team structure | VPE, CTO, CHRO, CFO |
PHASE 2: Independent Contributions (ISOLATED)
No cross-pollination. Each agent runs before seeing others' outputs.
Order: Research (if needed) → CMO → CFO → CEO → CTO → COO → CHRO → CRO → CISO → CPO → GC → CDO → CAIO → CCO → VPE (activated roles only)
Reasoning techniques: CEO: Tree of Thought (3 futures) | CFO: Chain of Thought (show the math) | CMO: Recursion of Thought (draft→critique→refine) | CPO: First Principles | CRO: Chain of Thought (pipeline math) | COO: Step by Step (process map) | CTO: ReAct (research→analyze→act) | CISO: Risk-Based (P×I) | CHRO: Empathy + Data | GC: Risk-Based (clause exposure) | CDO: Decision-Driven (what decision does this data drive) | CAIO: Eval-Demanding (no eval, no ship) | CCO: Retention-Obsessed (GRR over NRR) | VPE: Throughput-First (cycle-time math)
Contribution format (max 5 key points, self-verified):
## [ROLE] — [DATE]
Key points (max 5):
• [Finding] — [VERIFIED/ASSUMED] — ?/?/?
• [Finding] — [VERIFIED/ASSUMED] — ?/?/?
Recommendation: [clear position]
Confidence: High / Medium / Low
Source: [where the data came from]
What would change my mind: [specific condition]
Each agent self-verifies before contributing: source attribution, assumption audit, confidence scoring. No untagged claims.
PHASE 3: Critic Analysis
Executive Mentor receives ALL Phase 2 outputs simultaneously. Role: adversarial reviewer, not synthesizer.
Checklist:
- Where did agents agree too easily? (suspicious consensus = red flag)
- What assumptions are shared but unvalidated?
- Who is missing from the room? (customer voice? front-line ops?)
- What risk has nobody mentioned?
- Which agent operated outside their domain?
PHASE 4: Synthesis
Chief of Staff delivers using the Board Meeting Output format (defined in ../agent-protocol/SKILL.md):
- Decision Required (one sentence)
- Perspectives (one line per contributing role)
- Where They Agree / Where They Disagree
- Critic's View (the uncomfortable truth)
- Recommended Decision + Action Items (owners, deadlines)
- Your Call (options if founder disagrees)
PHASE 5: Human in the Loop ⏸️
Full stop. Wait for the founder.
⏸️ FOUNDER REVIEW — [Paste synthesis]
Options: ✅ Approve | ✏️ Modify | ❌ Reject | ❓ Ask follow-up
Rules:
- User corrections OVERRIDE agent proposals. No pushback. No "but the CFO said..."
- 30-min inactivity → auto-close as "pending review"
- Reopen any time with
/cs:boardroom resume
PHASE 6: Decision Extraction
After founder approval:
- Layer 1: Write full transcript →
~/.claude/decisions/raw/YYYY-MM-DD-<slug>.md - Layer 2: Write approved decision record →
~/.claude/decisions/approved/YYYY-MM-DD-<slug>.mdand append to the index~/.claude/decisions/approved/decisions.md - Mark rejected proposals
[DO_NOT_RESURFACE] - Confirm to founder with count of decisions logged, actions tracked, flags added
Memory Structure
Uses the canonical two-layer decision memory (see ../agent-protocol/SKILL.md → "Decision Memory (Canonical Layout)"):
~/.claude/decisions/
├── raw/YYYY-MM-DD-<slug>.md # Layer 1 — full transcripts (never auto-loaded)
├── raw/archive/YYYY/ # Raw transcripts after 90 days
├── approved/YYYY-MM-DD-<slug>.md # Layer 2 — founder-approved records (Phase 1 loads these)
└── approved/decisions.md # Layer 2 index — append-only
Future meetings load Layer 2 only. Never Layer 1. This prevents hallucinated consensus.
Migration: a legacy memory/board-meetings/ folder may exist from earlier versions; read it for history but write new transcripts and decisions to ~/.claude/decisions/.
Failure Mode Quick Reference
| Failure | Fix |
|---|---|
| Groupthink (all agree) | Re-run Phase 2 isolated; force "strongest argument against" |
| Analysis paralysis | Cap at 5 points; force recommendation even with Low confidence |
| Bikeshedding | Log as async action item; return to main agenda |
| Role bleed (CFO making product calls) | Critic flags; exclude from synthesis |
| Layer contamination | Phase 1 loads ~/.claude/decisions/approved/ only — hard rule |
References
templates/meeting-agenda.md— agenda formattemplates/meeting-minutes.md— final output formatreferences/meeting-facilitation.md— conflict handling, timing, failure modes
04 同仓库更多 Skills
webmcpify
Make a web app agent-ready — propose a WebMCP tool manifest, integrate, verify in a real browser, heal; unrelated code stays untouched. Use for "webmcpify", "add WebMCP", or "expose app actions to AI agents".
vardoger-analyze
Use when the user asks to personalize the GitHub Copilot CLI assistant, adapt Copilot to their style, use vardoger, or analyze their Copilot CLI conversation history. Reads the local session directory at `~/.copilot/session-state/`, extracts recurring preferences and conventions, and writes a fenced personalization block into `~/.copilot/copilot-instructions.md`. Runs entirely on the user's machine via the local `vardoger` CLI (`pipx install vardoger`); no network calls and no uploads. Triggers: 'personalize my copilot', 'analyze my copilot history', 'tailor copilot to me', 'run vardoger', 'update my copilot instructions from history', 'make copilot learn my style'.
transloadit-media-processing
Process media files (video, audio, images, documents) using Transloadit. Use when asked to encode video to HLS/MP4, generate thumbnails, resize or watermark images, extract audio, concatenate clips, add subtitles, OCR documents, or run any media processing pipeline. Covers 86+ processing robots for file transformation at scale.
terraform-azurerm-set-diff-analyzer
Analyze Terraform plan JSON output for AzureRM Provider to distinguish between false-positive diffs (order-only changes in Set-type attributes) and actual resource changes. Use when reviewing terraform plan output for Azure resources like Application Gateway, Load Balancer, Firewall, Front Door, NSG, and other resources with Set-type attributes that cause spurious diffs due to internal ordering changes.
technical-job-search
Use this skill when a software engineer asks for help with job search tasks: parsing or analyzing a job description, tailoring a CV/resume, writing a cover letter, evaluating a job offer, or drafting a post-interview follow-up email. Do not activate for general career advice unrelated to an active job search action.
steno-mode
Shorthand-first response compression that cuts ~40% of response tokens while preserving technical precision and exact literals. Use when the user says "steno mode", "shorthand mode", "compressed responses", "token reduction", "brief structured output", or invokes /steno. Supports four compression levels: lite, brief, court, machine. Do not trigger for requests needing polished prose such as onboarding/tutorial content, stakeholder or customer-facing copy, or teaching-focused explanations.