---
name: board-review
description: Conduct independent governance review of a material product proposal using customer, strategy, financial, technology, execution, trust, and risk perspectives. Use for high-impact, expensive, difficult-to-reverse, regulated, externally consequential, or disputed decisions requiring challenge, conditions, dissent, or escalation.
---

# Board Review

Challenge a material proposal independently. The board advises, conditions, rejects, or defers; it does not replace deterministic policy gates or required human authority.

## Materiality triggers

Invoke when one or more apply:

- major product, platform, or portfolio strategy change
- large investment or opportunity cost
- difficult-to-reverse architecture, contract, or ecosystem commitment
- significant customer, revenue, reputation, or operational impact
- privacy, security, safety, legal, ethical, or regulatory risk
- consequential public or partner commitment
- unresolved disagreement among accountable capabilities

Routine and reversible product work should not require a board review.

## Inputs

Prepare a common evidence packet using `references/board-review-template.md`:

- proposal and precise decision requested
- user and market evidence
- business, financial, and resource impact
- technical and operational evidence
- trust, security, privacy, legal, and compliance evidence
- alternatives, including doing nothing
- assumptions, missing evidence, and decision deadline

## Independence protocol

For material decisions, prefer separate agents for these perspectives:

- customer and product
- strategy and market
- financial and portfolio
- technology and execution
- trust and risk

Each director must receive the same evidence packet and form an opinion before seeing other opinions or the Product CEO's synthesis.

If isolated agents are unavailable:

1. disclose that independent review could not be performed
2. evaluate each perspective separately
3. label the result as a structured challenge, not an independent board
4. require human review for high-consequence decisions

## Director response

Each director returns:

- recommendation: Approve | Approve with conditions | Reject | Defer
- strongest supporting evidence
- strongest contrary evidence
- material risks
- binding or recommended conditions
- missing evidence
- confidence and rationale
- conflicts of interest or perspective limitations

## Chair synthesis

Only after all opinions are recorded:

1. compare recommendations, evidence, risk tolerance, and conditions
2. distinguish disagreement about facts from disagreement about values or appetite
3. identify deterministic policy, legal, security, privacy, or financial gates
4. issue one decision: Approve, Approve with conditions, Reject, or Defer
5. record binding conditions and owners
6. preserve minority opinions and material dissent
7. identify required human approvals
8. set a review date or evidence threshold for deferred decisions

A vote count is evidence of board sentiment, not a substitute for synthesis.

## Output

When authoring or revising a document, draft its substance, then invoke the bundled `classic-style` skill before final validation. Request **keep my structure; output only**. Preserve the required template, claims, citations, uncertainty, decisions, and code examples; polish prose only. Keep style notes in chat, outside the document. If skill invocation is unavailable, read `../classic-style/SKILL.md` and its reference and apply them directly. If those files are unavailable, report the missing dependency and ask to repair the installation rather than silently skipping it. Preserve each director's opinion and required confidence/verdict fields; a writing pass is not an independent board review.

When reviewing a feature specification, keep this board report separate from the spec. Return it to the orchestrator for a concise chat summary; do not append votes, conditions, dissent, or approval records to the product document. Only resulting product scope, experience constraints, or unresolved product questions may be reflected in the appropriate template sections.

Use `references/board-review-template.md` without omitting:

- evidence limitations
- each independent opinion
- chair decision
- binding conditions
- dissent
- independence limitations
- required human approvals
- review date

## Quality bar

- Directors challenge assumptions rather than repeat the proposal.
- Opinions cite the evidence packet and expose missing evidence.
- Material dissent remains visible.
- Conditions are testable and assigned.
- Deterministic restrictions cannot be overridden by agent consensus.
- The board never claims authority the human has not delegated.

## Boundaries

- Do not execute the proposal.
- Do not alter the evidence packet between directors.
- Do not expose private chain-of-thought; request concise rationale and evidence.
- Do not treat identical model outputs as proof of independent judgment.
- Escalate regulated, irreversible, destructive, or externally binding decisions to the human.