skills/used-car-investigator/references/decision-panel.md

# Ownership Panel — Deep Mode Only

Use this only for Deep reports or meaningful comparisons. The Standard dashboard already contains the
same decision factors and should omit this panel to save tokens. Use the panel as four analytical lenses,
not fictional people. Every lens uses the same cited evidence.
Do not simulate personalities, claim independent expert review, or decide by simple majority.

## Lenses

- **Mechanic — 30%:** failure modes, symptoms, maintenance, PPI targets, parts and labour exposure
- **Engineer — 25%:** architecture, thermal/lubrication design, durability, serviceability, repair paths
- **Financial — 25%:** price, depreciation, operating costs, insurance, repairs, downside and resale
- **Driver — 20%:** steering, ride, braking, performance, usability, efficiency, and buyer fit

History, title, structural safety, odometer integrity, and seller deception are shared hard-risk gates,
not a fifth opinion that can be outvoted. Adjust weights only when the buyer states priorities; disclose
the changed weights. A track-oriented buyer may raise Driver weight, but that cannot dilute safety or
catastrophic-risk gates.

## Output

Use a compact table:

| Lens | Call | Strongest reason | Condition or veto |
|---|---|---|---|

Calls are **Approve**, **Conditional**, or **Reject**. Then give one synthesized verdict using the
canonical 0-100 score. Panel calls explain tradeoffs; they do not create extra evidence or a second score.