# Lease Return

> A used electric car buyer's sitting in two steps: check the battery, then script the offer. Type what the listing and the car's own screen say; a window sticker of estimates (battery health, typical fade for the age and miles, range today, price per remaining kWh, battery warranty left) and an offer ladder run free in the browser. Two metered lanes write the battery check and the offer script. https://lease-return.skillsafe.ai/

## What it does

- Free, in the browser (leasekit.js, stickerview.js): battery health estimated from a battery health report, or from the range the car shows at a known charge divided by its range when new (rough - the range screen moves with recent driving and temperature). A typical fade for the car's age and miles (about 2.5 points in the first year, 1.6 a year after that, 0.6 per 10,000 miles, more in a hot climate or on mostly fast charging, less for LFP chemistry; a band of +/- 4 points), drawn as a fade chart with the 70% line many battery warranties use. Range today in mild weather, winter (about -30%) and at highway speed (about -15%); price per remaining kWh; the dollar value of any fade below typical at an estimated $140 per kWh; battery warranty left (8 years / 100,000 miles assumed unless typed). Listing words worth a question (salvage, rebuilt, flood, as-is, accident, recall, rideshare or fleet, warning lights, off-lease, one owner). A page read: go / inspect first / walk away.
- Free offer ladder: classic Ackerman steps (65, 85, 95, 100% of your target) or a softened set for car lots (85, 93, 97, 100%), the last step a precise non-round number just under the target, drawn as signposts on a road with your target, your walk-away and the asking price.
- Paid lane `inspect` (verdict go | inspect_first | walk_away): a battery read, a range read, red flags quoting the listing's own words, questions for the seller, test-drive checks, leverage for the offer, cautions.
- Paid lane `offer` (verdict offer | stretch | walk): an opening line that labels the seller's situation, an accusation audit, calibrated how/what questions, the page's four ladder steps each with words to say, one non-money ask, answers to likely pushback, a walk-away line.
- The page re-checks every reply: ladder amounts must match the page's ladder, percentages and dollar figures not on the sticker are named, red-flag words from the listing must be addressed, no internal field names in prose, verdicts compared with the page's own read.
- Exports: Markdown car file, JSON of every number, the offer ladder as CSV, a seller-question and test-drive checklist (.txt), the script to the clipboard. Saved cars live in the person's SkillSafe account (collection `cars`).

## Contract

- Run input: `task` (`inspect` | `offer`), `car`, `seller`, `page_numbers` (JSON), `page_flags` (JSON), `page_verdict`; `inspect` adds `listing`; `offer` adds `seller_said`, `asking`, `target`, `walk_away`, `ladder_style`, `page_ladder` (JSON), `battery_check`. See https://lease-return.skillsafe.ai/api.html
- Reply: one JSON object per task.

## Caveats

- Every number is an estimate from typed inputs and rule-of-thumb rates, not an inspection or a measurement. Get a battery health report and a pre-purchase inspection, and read your own warranty.

## Sources

- @personamanagmentlayer/automotive-expert (automotive systems, EV and battery management, diagnostics) - the battery check.
- @wondelai/negotiation (tactical empathy, labels, mirrors, calibrated questions, the Ackerman method) - the offer script.