Now liveThe Skillselion MCP - thousands of ranked skills, loaded into your agent mid-task. No install.Get it →
borghei avatar

Board Meeting

  • 83 installs
  • 451 repo stars
  • Updated July 21, 2026
  • borghei/claude-skills

board-meeting is a Claude skill that runs a structured 6-phase multi-agent deliberation to make strategic decisions while preventing groupthink.

About

board-meeting is a Claude skill that runs a structured multi-agent deliberation for strategic decisions. It executes six phases: context gathering, isolated C-suite contributions, critic analysis, synthesis, a full-stop founder review, and decision extraction. A founder uses it to make major strategic calls, resolve cross-functional disagreements, or evaluate irreversible choices while preventing groupthink.

  • Runs a structured 6-phase board-meeting deliberation for strategic decisions
  • Isolated C-suite contributions prevent groupthink and capture minority views
  • Full-stop founder review with a decision-extraction and logging phase

Board Meeting by the numbers

  • 83 all-time installs (skills.sh)
  • Ranked #1,431 of 3,282 Productivity & Planning skills by installs in the Skillselion catalog
  • Data as of Aug 5, 2026 (Skillselion catalog sync)
At a glance

board-meeting capabilities & compatibility

Free; guidance-based multi-agent protocol, no API keys stated.

Capabilities
board deck builder · strategic decision · multi agent deliberation
Use cases
planning · orchestration
Pricing
Free
From the docs

What board-meeting says it does

Multi-agent board meeting protocol for strategic decisions.
SKILL.md
Prevents groupthink through isolated contributions.
SKILL.md
No cross-pollination. Each advisor contributes without seeing others' outputs. This is the primary groupthink prevention mechanism.
SKILL.md
npx skills add https://github.com/borghei/claude-skills --skill board-meeting

Add your badge

Show developers this skill is listed on Skillselion. Paste this into your README.

Listed on Skillselion
Installs83
repo stars451
Last updatedJuly 21, 2026
Repositoryborghei/claude-skills

What it does

Run a structured 6-phase multi-agent board deliberation to reach a clean, logged strategic decision.

Who is it for?

Founders making major, hard-to-reverse strategic decisions who want multiple isolated perspectives.

When should I use this skill?

When making major strategic decisions, resolving cross-functional disagreements, or evaluating irreversible choices.

What you get

A synthesized, critic-reviewed decision with captured minority views and a logged rationale.

  • independent role contributions
  • critic analysis
  • synthesized decision and log

By the numbers

  • 6-phase protocol
  • maximum 6 roles per meeting
  • each advisor lists a maximum of 5 key points

Files

SKILL.mdMarkdownGitHub ↗

Board Meeting Protocol

Structured multi-agent deliberation that prevents groupthink, captures minority views, and produces clean, actionable decisions. Every phase has a purpose, a format, and rules that cannot be skipped.

Keywords

board meeting, executive deliberation, strategic decision, C-suite, multi-agent, founder review, decision extraction, independent perspectives, groupthink prevention, synthesis, critic analysis, structured deliberation

---

The 6-Phase Protocol

PHASE 1: Context Gathering
    |
PHASE 2: Independent Contributions (ISOLATED)
    |
PHASE 3: Critic Analysis (Executive Mentor)
    |
PHASE 4: Synthesis (Chief of Staff)
    |
PHASE 5: Founder Review (FULL STOP -- human decides)
    |
PHASE 6: Decision Extraction and Logging

---

Phase 1: Context Gathering

Purpose: Load all relevant context before anyone contributes.

Step 1: Load company context (if exists)
Step 2: Load decision history (Layer 2 ONLY -- NEVER raw transcripts)
Step 3: Reset session state -- no bleed from previous conversations
Step 4: Present agenda and activated roles
Step 5: Wait for founder confirmation before proceeding
Role Activation Matrix

Not all roles attend every meeting. Select based on topic:

Topic DomainActivateExclude
Market expansionCEO, CMO, CFO, CRO, COOCTO (unless tech expansion)
Product directionCEO, CPO, CTO, CMOCFO (unless budget question)
Hiring / orgCEO, CHRO, CFO, COOCMO, CTO (unless their teams)
PricingCMO, CFO, CRO, CPOCTO, CHRO
TechnologyCTO, CPO, CFO, CISOCMO, CRO
FundraisingCEO, CFO, CROCISO, CHRO
Security incidentCEO, CTO, CISO, COOCMO, CRO
M&ACEO, CFO, CTO, CHRO, COO-- (all relevant)

Maximum attendees: 6 roles per meeting. More than 6 creates noise, not insight.

---

Phase 2: Independent Contributions (ISOLATED)

Critical Rule: No cross-pollination. Each advisor contributes without seeing others' outputs. This is the primary groupthink prevention mechanism.

Contribution Order
1. Research/data gathering (if needed)
2. CMO  -- market perspective
3. CFO  -- financial perspective
4. CEO  -- strategic perspective
5. CTO  -- technical perspective
6. COO  -- operational perspective
7. CHRO -- people perspective
8. CRO  -- revenue perspective
9. CISO -- security/risk perspective
10. CPO -- product perspective
Contribution Format (Strict)

Each advisor's contribution must follow this exact format:

## [ROLE] -- [DATE]

Key Points (maximum 5):
  1. [Finding] -- Confidence: [High/Medium/Low] -- Source: [data source]
  2. [Finding] -- Confidence: [High/Medium/Low] -- Source: [data source]
  3. [Finding] -- Confidence: [High/Medium/Low] -- Source: [data source]

Recommendation: [Clear position statement]
Confidence: [High / Medium / Low]
Key Assumption: [The one assumption this recommendation depends on]
What Would Change My Mind: [Specific condition or data point]
Reasoning Techniques by Role
RoleTechniqueHow It Works
CEOTree of ThoughtExplore 3 possible futures, evaluate each
CFOChain of ThoughtShow the math, step by step
CMORecursion of ThoughtDraft -> self-critique -> refine
CPOFirst PrinciplesDecompose to fundamental user needs
CROChain of ThoughtPipeline math must be explicit
COOStep by StepMap the operational process
CTOAnalyze then ActResearch -> analyze -> recommend
CISORisk-BasedProbability x Impact for every option
CHROEmpathy + DataHuman impact first, then validate with metrics

---

Phase 3: Critic Analysis

Purpose: The Executive Mentor receives ALL Phase 2 outputs simultaneously and performs adversarial review.

Critic Checklist
CheckQuestion
Suspicious consensusWhere did agents agree too easily?
Shared assumptionsWhat assumptions are shared but unvalidated?
Missing voiceWho is not in the room? (customer voice? front-line ops?)
Unmentioned riskWhat risk has nobody mentioned?
Domain bleedDid any agent operate outside their domain?
Data qualityWhich claims are backed by data vs. assumption?
ReversibilityHas anyone assessed if this decision can be undone?
Critic Output Format
## CRITIC ANALYSIS

Consensus Assessment:
  [Genuine agreement / Suspicious alignment / Split decision]

Unvalidated Assumptions:
  1. [Assumption shared by multiple advisors but not verified]
  2. [Assumption]

Missing Perspectives:
  - [Voice or data point not represented]

Unmentioned Risks:
  - [Risk nobody raised]

Domain Violations:
  - [If any agent operated outside their domain]

The Uncomfortable Truth:
  [The one thing nobody wants to say but needs to be said]

---

Phase 4: Synthesis

Purpose: Chief of Staff combines all inputs into a decision-ready format.

Synthesis Structure
## BOARD MEETING SYNTHESIS
Topic: [topic]
Date: [date]
Attendees: [roles]

### Decision Required
[One sentence: what must be decided]

### Perspectives Summary
| Role | Position | Confidence | Key Concern |
|------|----------|-----------|-------------|
| [Role] | [1-line summary] | [H/M/L] | [Top concern] |
| [Role] | [1-line summary] | [H/M/L] | [Top concern] |

### Where They Agree
[2-3 consensus points]

### Where They Disagree
[Named conflicts with each side's reasoning]
[What the disagreement is really about]

### Critic's View
[The uncomfortable truth from Phase 3]

### Recommended Decision
[Clear recommendation with rationale]

### Action Items (if approved)
1. [Action] -- Owner: [role] -- Deadline: [date]
2. [Action] -- Owner: [role] -- Deadline: [date]
3. [Action] -- Owner: [role] -- Deadline: [date]

### Your Call
[If you disagree with the recommendation, here are alternatives:]
Option A: [description] -- Trade-off: [what you gain/lose]
Option B: [description] -- Trade-off: [what you gain/lose]

---

Phase 5: Founder Review

FULL STOP. Wait for the founder. No agent acts beyond this point.

FOUNDER REVIEW

[Paste synthesis above]

Options:
  [A] Approve as recommended
  [M] Modify (specify changes)
  [R] Reject (specify reason)
  [Q] Ask follow-up question to specific role
  [D] Defer decision (specify timeline)
Phase 5 Rules
RuleRationale
Founder corrections override all agent proposalsHuman judgment is final
No pushback on founder decisionsAgents advise, founder decides
30-minute inactivity auto-closes as "pending review"Prevents zombie meetings
Founder can reopen any timeDecisions are not time-locked
Follow-up questions go to specific roleKeeps discussion focused

---

Phase 6: Decision Extraction

Purpose: After founder approval, extract and log all decisions.

Step 1: Write full transcript to Layer 1
  --> memory/board-meetings/YYYY-MM-DD-raw.md

Step 2: Run conflict detection against existing decisions
  --> Check for DO_NOT_RESURFACE violations
  --> Check for topic contradictions
  --> Check for owner conflicts

Step 3: Surface any conflicts to founder for resolution

Step 4: Append approved decisions to Layer 2
  --> memory/board-meetings/decisions.md

Step 5: Mark rejected proposals with DO_NOT_RESURFACE

Step 6: Confirm to founder:
  "Meeting concluded. Logged: [N] decisions, [M] action items,
   [K] DO_NOT_RESURFACE flags."

---

Failure Mode Reference

FailureDetectionFix
GroupthinkAll advisors agree without tensionRe-run Phase 2 isolated; force "strongest argument against"
Analysis paralysisDiscussion exceeds 5 points per advisorCap at 5; force recommendation even with Low confidence
BikesheddingDiscussion on minor points, major decisions deferredLog as async action; return to main agenda
Role bleedCFO making product calls, CTO pricingCritic flags in Phase 3; exclude from synthesis
Layer contaminationRaw transcripts loaded in Phase 1Hard rule: decisions.md only. Never raw.
Founder absencePhase 5 timeoutAuto-close as pending. No decisions without founder.
Stale contextCompany context not loadedPhase 1 mandatory context check
Missing roleKey perspective not activatedChief of Staff reviews topic against routing matrix

---

Meeting Cadence

TriggerMeeting TypeTypical Duration
Scheduled quarterlyFull strategic review2-3 hours
Complexity score >= 8On-demand strategic1-2 hours
Cross-functional conflictResolution meeting1 hour
Crisis or urgent decisionEmergency session30-60 minutes
Founder requestAny topicVaries

---

Red Flags

  • Board meetings consistently produce no decisions -- meeting is theater
  • Same topic discussed in 3+ meetings -- decision avoidance
  • Phase 2 contributions all align perfectly -- isolation was breached or topic is trivial
  • No Phase 3 (critic) conducted -- groupthink risk
  • Founder skipping Phase 5 -- decisions without accountability
  • Decisions logged but never reviewed -- decision logger not functioning
  • Meeting attendees always include all roles -- topic selection not working

---

Output Artifacts

RequestDeliverable
"Convene the board on [topic]"Full 6-phase protocol execution
"Quick advisory meeting"Abbreviated: Phase 1-2-4-5 (skip critic)
"Review a past meeting"Load Layer 1 raw transcript (explicit request only)
"What did we decide about [topic]?"Search Layer 2 decision history
"Resume a pending meeting"Reload Phase 5 with pending synthesis

---

Tool Reference

meeting_simulator.py

Validates role activation, contribution completeness, and phase sequencing.

# Simulate with defaults
python scripts/meeting_simulator.py

# Specify topic and complexity
python scripts/meeting_simulator.py --topic "Series B timing" --type fundraising --complexity 9

# Specify activated roles
python scripts/meeting_simulator.py --type m_and_a --roles CEO CFO CTO CHRO

# List all topic types and required roles
python scripts/meeting_simulator.py --list-topics

# JSON output
python scripts/meeting_simulator.py --type strategy --json

decision_tracker.py

Tracks board decisions, detects conflicts, flags overdue reviews and actions.

# Track demo decisions
python scripts/decision_tracker.py

# From decision log file
python scripts/decision_tracker.py --input decisions.json

# JSON output
python scripts/decision_tracker.py --json

complexity_scorer.py

Scores decision complexity to determine single/dual/multi-advisor or board routing.

# Score with CLI flags
python scripts/complexity_scorer.py --topic "Market expansion" --domains 2 --reversibility 2 --financial 1 --team 2 --urgency 0

# Add modifiers
python scripts/complexity_scorer.py --topic "Acquisition" --domains 2 --reversibility 2 --financial 2 --team 2 --urgency 1 --modifiers cross_functional external_stakeholders sets_precedent

# JSON output
python scripts/complexity_scorer.py --topic "Pricing change" --json

---

Troubleshooting

ProblemLikely CauseFix
All advisors agree without any tension in Phase 2Groupthink or trivial topic; isolation may have been breachedRe-run Phase 2 with forced "strongest argument against" from each role
Discussion exceeds 5 points per advisorAnalysis paralysis; no cap enforcedHard cap at 5 key points; force a recommendation even with Low confidence
Phase 5 times out with no founder responseFounder absence or decision avoidanceAuto-close as "pending review" at 30 min; no decisions without founder
Same topic discussed in 3+ meetingsDecision avoidance or new data not surfacedEscalate: force decision or formally defer with stated timeline
Decisions logged but never reviewedDecision logger not integrated into meeting cadenceAdd "previous decisions review" to Phase 1 context loading
Roles operating outside their domainNo critic analysis conducted or critic missed itEnforce Phase 3 critic checklist; flag domain violations explicitly

---

Success Criteria

  • Every board meeting produces at least 1 logged decision with owner, deadline, and review date
  • Phase 2 contributions are independently generated (zero cross-pollination incidents per quarter)
  • Phase 3 critic analysis identifies at least 1 unvalidated assumption per meeting
  • Founder approval/modification/rejection captured within 30 minutes of synthesis presentation
  • Decision history has zero conflicting active decisions (conflicts detected and resolved)
  • Meeting duration stays within 2 hours for standard strategic reviews, 1 hour for resolution meetings
  • 90%+ of logged decisions have action items completed by their stated deadlines

---

Scope & Limitations

In Scope: Multi-agent deliberation protocol, role activation matrix, contribution formats, critic analysis, synthesis, decision extraction, decision conflict detection, meeting simulation.

Out of Scope: Actual AI agent orchestration (this is a protocol specification, not runtime code), real-time meeting facilitation, video/audio recording, external board member management.

Limitations: The protocol assumes all advisor contributions are available in text format. Complexity scoring provides routing guidance but cannot account for political dynamics. Decision conflict detection works on exact topic matching -- semantic conflicts across different topics require human judgment.

---

Integration Points

SkillIntegration
chief-of-staffRoutes questions that score 9-10 complexity into the board meeting protocol
decision-loggerPhase 6 feeds decisions directly into the two-layer decision memory
board-deck-builderBoard deck sections provide pre-read context for Phase 1
executive-mentorPhase 3 critic analysis can be performed by the Executive Mentor skill
ceo-advisor through ciso-advisorAll C-suite advisors contribute independently in Phase 2
strategic-alignmentValidates that meeting decisions align with strategic goals

Related skills

FAQ

What are the six phases?

Context gathering, independent (isolated) contributions, critic analysis, synthesis, founder review (full stop), and decision extraction and logging.

How does it prevent groupthink?

Each advisor contributes without seeing others' outputs, so there is no cross-pollination during Phase 2.

This week in AI coding

Five minutes, every Monday - the tools, releases and tactics for developers.

unsubscribe anytime.