/mirror

Analyze blind spots in a decision or situation. Surface what you might be missing.

Phase 1: Context Capture

State the situation clearly:

• Core decision or situation
• Current framing
• Stakeholders involved
• Timeline/urgency

Phase 2: Blind Spot Detection

Examine 5 categories systematically:

1. Assumptions Audit

• What am I treating as fact without verification?
• Which "always" or "never" statements exist?
• What would someone from outside my field question immediately?

2. Incentive Analysis

• Who benefits if this succeeds?
• Who benefits if this fails?
• What incentives am I overlooking?
• What conflicts of interest exist?

3. Missing Perspectives

• Who's not in the room but should be?
• What expertise is absent?
• Which user segments am I ignoring?
• What would [specific role] say about this?

4. Second-Order Effects

• What happens after the immediate consequence?
• Which systems interact with this decision?
• What precedent does this set?
• What becomes easier/harder next?

5. Emotion vs Logic

• What am I avoiding out of discomfort?
• Which decisions feel "obvious" (beware)?
• What would I do if fear/ego were removed?
• Am I solving the stated problem or my anxiety about it?

Phase 3: Red Team

Play devil's advocate:

• Steel-man the opposite decision
• List 3 ways this could fail spectacularly
• Identify what would make you reverse this decision in 6 months

Phase 4: Synthesis

Present findings:

## Blind Spots Detected

### Critical (address before proceeding)
- [Blind spot]: [Why it matters] - [How to verify]

### Significant (monitor closely)
- [Blind spot]: [Why it matters] - [Mitigation]

### Minor (acknowledge)
- [Blind spot]: [Why it matters]

## Questions to Answer
1. [Question that would resolve uncertainty]
2. [Question that tests assumption]
3. [Question that surfaces risk]

---
[Quality Footer - see ../skills/soul/references/quality-footer.md]

Phase 5: Wait

Do not recommend action. Present blind spots and wait for user direction.

---

Task

$ARGUMENTS