Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Design a fair, consistent tenant screening process for a rental. Use when asked how to screen tenants, set rental criteria, evaluate rental applicants, or build a tenant screening process. Produces a screening framework — written objective criteria, the application & checks, a consistent evaluation method, and applicant communication — built to be fair and Fair-Housing-compliant. Not legal advice.
.claude/skills/mohitagw15856-tenant-screening-guide/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-01 | ✗→✓ | ▲ Improved | 24% | 0% |
| case-02 | ✗→✓ | ▲ Improved | 5% | 0% |
| case-03 | ✗→✓ | ▲ Improved | 9% | 0% |
| case-07 | ✗→✓ | ▲ Improved | 21% | 0% |
| case-08 | ✗→✓ | ▲ Improved | 51% | 0% |
Good tenant screening is consistent and criteria-based: the same written standards applied to every applicant, judged on objective, rental-relevant factors. That protects the landlord (better tenants, fewer problems) and keeps the process fair and legal. This skill builds that framework — the criteria, the checks, and a consistent way to decide — so screening isn't ad-hoc or discriminatory.
> Note: this is a process aid, not legal advice. Tenant screening is heavily regulated — Fair Housing > laws (protected classes), FCRA/background-check rules, source-of-income and criminal-history limits, and > local ordinances vary widely and change. Apply criteria identically to all applicants, and have your > criteria and process reviewed by a qualified attorney/property manager for your jurisdiction.
Given "help me screen tenants for my rental", produce the framework anyway — propose objective, rental-relevant criteria and a consistent process, clearly flagging every legally-sensitive choice (confirm with local law/attorney). Never propose criteria based on protected characteristics; emphasise consistency.
Ask for these only if they aren't already provided (else use labelled defaults):
1. Written objective criteria — the standards applied to every applicant, e.g.:
Each marked (confirm against local law) where sensitive.
2. Application & checks — what to collect (application form, ID, income proof, references) and the checks (credit/background) with required applicant consent (FCRA).
3. Consistent evaluation — apply the criteria the same way to all applicants; ideally first-qualified-first or a scored checklist — documented, so decisions are defensible.
4. Applicant communication — clear criteria up front, and proper adverse-action notice if you decline based on a report (an FCRA requirement) — flagged to confirm.
5. Compliance guardrails — apply identically to everyone; judge only rental-relevant, objective factors; never screen or comment on protected characteristics (race, colour, religion, sex, familial status, national origin, disability, and other protected classes); respect source-of-income and criminal-history limits where they apply.
Add a prominent note to have the framework reviewed by a local attorney/property manager.
Fair-housing & tenant-screening practice — written objective criteria applied consistently, FCRA-compliant checks and notices, and protected-class safeguards (jurisdiction review required).
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→pass | 26,927 | 27,578 | +2% | 1 | 1 | 0% | 3,827 | 4,739 | +24% | 0 | 0 | — |
case-02 | fail→pass | 34,429 | 28,187 | -18% | 1 | 1 | 0% | 4,096 | 4,313 | +5% | 0 | 0 | — |
case-03 | fail→pass | 34,281 | 30,019 | -12% | 1 | 1 | 0% | 4,019 | 4,394 | +9% | 0 | 0 | — |
case-04 | pass→pass | 21,148 | 20,037 | -5% | 1 | 1 | 0% | 2,528 | 2,831 | +12% | 0 | 0 | — |
case-05 | pass→pass | 25,575 | 22,556 | -12% | 1 | 1 | 0% | 2,929 | 3,809 | +30% | 0 | 0 | — |
case-06 | pass→pass | 23,208 | 27,051 | +17% | 1 | 1 | 0% | 2,781 | 4,112 | +48% | 0 | 0 | — |
case-07 | fail→pass | 23,869 | 15,444 | -35% | 1 | 1 | 0% | 2,591 | 3,142 | +21% | 0 | 0 | — |
case-08 | fail→pass | 25,349 | 25,506 | +1% | 1 | 1 | 0% | 2,491 | 3,773 | +51% | 0 | 0 | — |
case-09 | fail→pass | 24,399 | 23,269 | -5% | 1 | 1 | 0% | 2,849 | 3,960 | +39% | 0 | 0 | — |
case-10 | fail→pass | 18,042 | 13,450 | -25% | 1 | 1 | 0% | 2,142 | 3,264 | +52% | 0 | 0 | — |
case-11 | fail→pass | 33,490 | 19,376 | -42% | 1 | 1 | 0% | 2,799 | 3,724 | +33% | 0 | 0 | — |
case-12 | pass→fail | 19,805 | 22,651 | +14% | 1 | 1 | 0% | 2,535 | 3,411 | +35% | 0 | 0 | — |
case-13 | fail→pass | 23,065 | 27,103 | +18% | 1 | 1 | 0% | 2,339 | 3,844 | +64% | 0 | 0 | — |
case-14 | fail→fail | 21,703 | 19,585 | -10% | 1 | 1 | 0% | 1,321 | 4,062 | +207% | 0 | 0 | — |
case-15 | fail→pass | 21,844 | 25,399 | +16% | 1 | 1 | 0% | 2,644 | 3,680 | +39% | 0 | 0 | — |
case-16 | fail→pass | 13,174 | 20,863 | +58% | 1 | 1 | 0% | 1,718 | 3,629 | +111% | 0 | 0 | — |
case-17 | fail→pass | 16,474 | 21,029 | +28% | 1 | 1 | 0% | 2,462 | 3,913 | +59% | 0 | 0 | — |
case-18 | fail→pass | 17,239 | 22,423 | +30% | 1 | 1 | 0% | 1,896 | 3,761 | +98% | 0 | 0 | — |
case-19 | pass→pass | 25,043 | 14,751 | -41% | 1 | 1 | 0% | 2,551 | 3,280 | +29% | 0 | 0 | — |
case-20 | fail→pass | 14,344 | 22,690 | +58% | 1 | 1 | 0% | 1,282 | 3,292 | +157% | 0 | 0 | — |
case-21 | fail→pass | 22,487 | 26,035 | +16% | 1 | 1 | 0% | 2,556 | 4,233 | +66% | 0 | 0 | — |
case-22 | pass→pass | 25,373 | 16,387 | -35% | 1 | 1 | 0% | 2,633 | 3,467 | +32% | 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. 22 cases were attempted. The headline lift of +64 percentage points is the difference between those two pass rates over the 22 comparable cases. 1 case got worse with the skill loaded, and it is included in that figure.
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.