Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Get your security deposit back — the move-out documentation that wins disputes before they start, the itemized-deduction challenge, the demand-letter ladder, and the small-claims decision point. Use when asked how do I get my deposit back, my landlord is keeping my deposit, dispute these deposit deductions, or write a deposit demand letter. Produces the move-out evidence protocol, the deduction-by-deduction challenge with the wear-and-tear line drawn, the escalation ladder with letters, and the
.claude/skills/mohitagw15856-security-deposit-recovery/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-02 | ✗→✓ | ▲ Improved | 2% | 0% |
| case-12 | ✗→✓ | ▲ Improved | 77% | 0% |
| case-06 | ✓→✓ | = Same ✓ | 92% | 0% |
| case-19 | ✓→✓ | = Same ✓ | 60% | 0% |
| case-04 | ✓→✓ | = Same ✓ | 82% | 0% |
Deposit disputes are won at move-out, weeks before the landlord decides anything: the tenant with timestamped photos of every wall, a completed walkthrough, and a forwarding address in writing collects; the tenant with memories negotiates. This skill runs both phases — the evidence protocol while there's still access, and the challenge-and-escalate ladder when deductions arrive — anchored on the distinction that decides almost every dispute: normal wear and tear (the landlord's cost of doing business, in most jurisdictions not deductible) versus damage (yours).
Ask for these if not provided:
The sweep checklist by room · the walkthrough request wording · receipts to keep · the forwarding-address letter, dated]
| Claimed deduction | Amount | Wear / damage / unsubstantiated | The response (with depreciation math where it applies) | |---|---|---|---|
Each rung with its letter drafted verbatim, its deadline, and the elapsed-time citation where the clock has run · the multiplied-damages category flagged verify-locally]
The economics (filing cost vs. amount) · the evidence binder order · how the hearing runs · the settle-on-the-courthouse-steps pattern to expect]
> Deposit deadlines, deduction rules, and penalty provisions are jurisdiction-specific — verify the local specifics before citing exact numbers; the letters here cite rule categories for exactly that reason. Not legal advice.
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-06 | pass→pass | 20,842 | 32,109 | +54% | 1 | 1 | 0% | 2,442 | 4,678 | +92% | 0 | 0 | — |
case-19 | pass→pass | 23,019 | 25,492 | +11% | 1 | 1 | 0% | 2,513 | 4,030 | +60% | 0 | 0 | — |
case-01 | fail→fail | 37,017 | 37,505 | +1% | 1 | 1 | 0% | 5,438 | 6,334 | +16% | 0 | 0 | — |
case-02 | fail→pass | 48,119 | 43,774 | -9% | 1 | 1 | 0% | 6,914 | 7,070 | +2% | 0 | 0 | — |
case-03 | fail→fail | 41,475 | 35,310 | -15% | 1 | 1 | 0% | 6,304 | 5,621 | -11% | 0 | 0 | — |
case-04 | pass→pass | 20,417 | 17,284 | -15% | 1 | 1 | 0% | 2,138 | 3,900 | +82% | 0 | 0 | — |
case-05 | pass→pass | 15,018 | 18,417 | +23% | 1 | 1 | 0% | 1,680 | 3,387 | +102% | 0 | 0 | — |
case-07 | fail→fail | 16,661 | 27,731 | +66% | 1 | 1 | 0% | 2,353 | 4,680 | +99% | 0 | 0 | — |
case-08 | pass→pass | 14,296 | 28,785 | +101% | 1 | 1 | 0% | 2,257 | 4,798 | +113% | 0 | 0 | — |
case-09 | pass→pass | 20,378 | 26,343 | +29% | 1 | 1 | 0% | 2,495 | 4,645 | +86% | 0 | 0 | — |
case-10 | pass→pass | 17,450 | 25,028 | +43% | 1 | 1 | 0% | 2,316 | 4,298 | +86% | 0 | 0 | — |
case-11 | fail→fail | 14,010 | 20,786 | +48% | 1 | 1 | 0% | 2,359 | 4,105 | +74% | 0 | 0 | — |
case-12 | fail→pass | 20,639 | 27,213 | +32% | 1 | 1 | 0% | 2,386 | 4,229 | +77% | 0 | 0 | — |
case-13 | pass→pass | 12,717 | 46,354 | +265% | 1 | 1 | 0% | 2,160 | 4,794 | +122% | 0 | 0 | — |
case-14 | pass→pass | 19,306 | 22,535 | +17% | 1 | 1 | 0% | 2,323 | 4,758 | +105% | 0 | 0 | — |
case-15 | pass→pass | 28,052 | 22,628 | -19% | 1 | 1 | 0% | 2,957 | 4,051 | +37% | 0 | 0 | — |
case-16 | pass→pass | 11,964 | 23,765 | +99% | 1 | 1 | 0% | 1,670 | 3,772 | +126% | 0 | 0 | — |
case-17 | pass→pass | 12,627 | 24,715 | +96% | 1 | 1 | 0% | 1,794 | 3,846 | +114% | 0 | 0 | — |
case-18 | pass→pass | 10,452 | 17,724 | +70% | 1 | 1 | 0% | 1,593 | 3,107 | +95% | 0 | 0 | — |
case-20 | pass→pass | 26,143 | 34,606 | +32% | 1 | 1 | 0% | 2,905 | 5,335 | +84% | 0 | 0 | — |
case-21 | pass→pass | 19,822 | 30,215 | +52% | 1 | 1 | 0% | 2,288 | 4,451 | +95% | 0 | 0 | — |
case-22 | pass→pass | 22,715 | 126,057 | +455% | 1 | 1 | 0% | 2,285 | 4,287 | +88% | 0 | 0 | — |
case-23 | pass→pass | 22,039 | 30,033 | +36% | 1 | 1 | 0% | 2,224 | 4,501 | +102% | 0 | 0 | — |
DecimalAI ran this skill against gemini-3.6-flash twice over the same eval suite — once with the skill loaded and once without — and compared the two runs case by case. 23 cases were attempted. The headline lift of +9 percentage points is the difference between those two pass rates over the 23 comparable cases.
Without the skill loaded, the model failed this case. With it loaded, the same prompt on the same model passed. This is one improved case from the latest verified run; every case, including any that regressed, is in the table above.
Other measured skills in the registry, with their headline benchmark lift.