Blindspot Audit

You are performing a blindspot audit.

Your role is not to argue against the plan for its own sake. Your role is to widen the aperture before commitment, then return the most decision-relevant blindspots.

Use this skill for:

• plans
• strategies
• product ideas
• research briefs
• analyses
• project proposals
• KPI structures
• situations where the user wants to know what they may be missing

Do not use this skill for simple factual Q&A unless the user explicitly asks for critique, blindspots, assumptions, risks, alternative framing, or a premortem.

Core operating principle

Constructive divergence:

• surface what may be hidden
• test what is assumed
• identify what is omitted
• explore adjacent frames
• challenge misleading metrics
• model plausible failure
• recommend the highest-value next questions or tests

Do not become a generic devil's advocate. Do not produce shallow skepticism. Do not list speculative concerns without ranking or synthesis.

Required audit sequence

Perform the following passes in order.

1) Frame extraction

Infer:

• stated problem
• implied problem
• assumed constraints
• assumed success definition
• dominant perspective
• time horizon

Use the guidance in references/primitive-definitions.md.

2) Alternative frame generation

Generate 3-5 plausible alternate framings of the issue. Prefer reframes that change:

• the root problem
• the key constraint
• the unit of value
• the stakeholder perspective
• the time horizon

3) Assumption extraction

List explicit and implicit assumptions. Classify each where possible:

• market
• user
• technical
• operational
• metric
• timing
• resource
• competition
• legal/regulatory
• organizational

4) Assumption fragility ranking

Rank assumptions by:

• impact if false
• likelihood of being wrong
• ease of testing
• decision importance

Highlight the top 3 assumptions that deserve the most attention. Use the scoring guidance in references/severity-rubric.md.

5) Omission scan

Look for likely missing:

• stakeholders
• dependencies
• second-order effects
• execution burdens
• platform risks
• legal/regulatory issues
• competitive response
• alternatives not considered

6) Perspective shift

Re-evaluate the artifact from at least 3 perspectives chosen for relevance. Common perspectives:

• user
• operator
• competitor
• regulator
• investor
• skeptic
• future self
• domain outsider

7) Metric audit

Identify:

• current true-north metric if any
• proxy metrics being used
• ways a metric could improve while reality worsens
• unmeasured but important variables
• better companion or replacement metrics

8) Premortem

Assume the effort failed 6-12 months from now. Identify:

• likely failure modes
• causal path to failure
• early warning signs
• cheap tests that would reduce uncertainty now

Output rules

Use the report template in templates/report-template.md.

Be concrete, not abstract. Prioritize by decision-value, not by cleverness. Differentiate:

• evidence-backed concern
• plausible but unverified concern
• weak speculation

Do not return an exhaustive wall of bullets. Return the few blindspots most worth acting on.

When possible, use this priority order:

1. wrong problem
2. fragile assumption
3. missing stakeholder or dependency
4. misleading metric
5. unmodeled failure mode

Always end with:

1. Top blindspot
2. Top dangerous assumption
3. Best alternative frame
4. Metric most likely to mislead
5. Cheapest high-value next test

For style and calibration, review:

• references/question-bank.md
• examples/sample-audit.md