Best Claude Prompts for Real Estate Agents in 2026

Three properties separate useful prompts from liability prompts. **Fair Housing discipline:** no preferential language about race, color, national origin, religion, sex (including sexual orientation and gender identity per HUD's 2021 directive), familial status, or disability — including proxies like "perfect for a young professional" or "walk to church". **Source grounding:** stats come from MLS, Zillow, Redfin, or RealTrends data you supply, not Claude's training data. **Broker-review assumption:** every commission, buyer-rep, and HOA paragraph is draft input to your managing broker.

Per Fair Housing Act §3604(c), publishing any ad indicating protected-class preference is unlawful — liability is strict, intent does not matter. Per the NAR settlement changes, MLS Participants working with buyers must sign a written buyer-rep before touring, with compensation in objective terms.

**Prompt block:** "You are a listing copywriter. I'll paste property notes plus the MLS sheet. Produce a 120-word public-remarks paragraph and 60-word agent-remarks paragraph. Hard constraints: describe the property only; no protected-class references or proxies ('family friendly', 'walk to church', 'safe neighborhood', 'great schools' without a source, 'master bedroom' — use 'primary bedroom'); quantitative claims must match what I supplied; flag Fair Housing risks with [FH REVIEW]."

**Fair Housing flag — HIGH.** Per HUD FHEO guidance, "family room" describing a room is fine; "perfect for families" is not. NFHA 2024 logged record complaints tied to discriminatory marketing.

**Sample output:** "Three-bed, two-bath ranch on a 0.28-acre corner lot. Updated kitchen (2023), quartz counters, stainless. Refinished hardwoods. Primary bedroom en-suite + walk-in. New roof (2024), tankless water heater, two-car garage. [FH REVIEW] 'quiet street' → 'cul-de-sac location'."

**Prompt block:** "I'll paste verified market data for [NEIGHBORHOOD] from MLS, Zillow, Redfin, and the RealTrends quarterly. Write a 250-word neighborhood narrative for a buyer. Use only the stats supplied — cite the source for each. Cover inventory, DOM, list-to-sale, price trend. Do not describe the people. Do not characterize schools beyond the GreatSchools rating I provided. No 'safe', 'family friendly', 'desirable demographics'. End with three questions the buyer should answer before writing an offer."

**Fair Housing flag — HIGH.** Second-highest exposure surface after listing copy. Any language signaling who "belongs" can violate §3604(c).

**Sample output:** "Maple Heights closed May at 1.4 months supply (MLS), down from 2.1 YoY. Median DOM 11 (Redfin). List-to-sale 102.3% (MLS YTD). Median sold $487K (Zillow, 90d), +6.2% YoY. Per RealTrends Q1 2026, metro is at 2.6 months — Maple Heights runs hotter than the region."

**Prompt block:** "I'll paste a 6-comp CMA (active, pending, sold) and the absorption rate. Draft a 1-page CMA narrative letter. Sections: market context (cite my data; no invented stats), pricing strategy tied to comps, 14-day re-evaluation trigger, what we control vs. what the market controls. No price or timeline guarantees. No protected-class buyer descriptors. End with a clear next-step ask."

**Fair Housing flag — LOW.** Seller-facing copy carries less §3604(c) exposure. **Disclosure flag:** compensation references must comply with state advertising rules and the post-settlement disclosure standards.

**Sample output:** "List band: $462K–$478K. Three closed comps within 0.4 mi (avg $471/sqft). Active at $499K has 38 DOM — above absorption. Day-14 trigger: <8 showings or 0 offers → revisit."

**Prompt block:** "I'll provide list price, offer terms, DOM, neighborhood absorption (months of supply), list-to-sale ratio. Output: market posture in one sentence (seller's / balanced / buyer's with the absorption threshold cited), two counter scenarios tied to absorption + DOM, non-price levers worth 0.5-1.5% (close date, leaseback, repair credit, appliances, inspection scope), one walk-away signal. No appraisal prediction or guarantees."

**Fair Housing flag — LOW.** Negotiation framing rarely creates protected-class exposure. **Practice flag:** absorption thresholds vary by market; use NAR Local Market Reports or your MLS's months-of-supply calc.

**Sample output:** "Seller's market — 1.6 months supply (NAR threshold <6). Counter A: hold list, shorten inspection to 5d, close in 21d. Counter B: drop $5K, request appliance retention ~$3,500."

**Prompt block:** "You are explaining the written buyer-rep agreement to a first-time buyer. Constraints: the August 17, 2024 NAR settlement practice changes require a signed buyer-rep before touring an MLS-listed home; the agreement must state compensation as a specific amount or rate in objective terms — never 'whatever the seller offers'; compensation can be paid by seller, listing broker, buyer directly, or a combination, disclosed in writing; do not give legal advice; flag any sentence requiring broker review with [BROKER REVIEW]. Write a 350-word plain-English explainer at 8th-grade reading level. End with: 'I will send you the form and review it with you before you sign anything.'"

**Compliance flag — VERY HIGH.** Most-changed area post-settlement. Exact language, timing, and the objective-terms rule are enforceable with state license boards. Broker review of every explainer is non-negotiable.

**Sample output:** "A buyer-rep agreement is a written contract between you and me. After August 17, 2024, the NAR settlement says we sign one before I show any MLS-listed home. It spells out what I do and how I get paid — a specific dollar amount or percentage, not 'whatever the seller offers'. [BROKER REVIEW] Compensation can come from the seller, listing broker, you directly, or a mix."

**Prompt block:** "Draft a 4-message FSBO sequence (Day 1 SMS, Day 3 email, Day 7 voicemail, Day 14 note). Constraints: TCPA consent + opt-out; CAN-SPAM ad ID + physical address + 10-day opt-out; lead with a market data point sourced from MLS, Zillow, or Redfin; offer value before any ask; never imply MLS access if the FSBO is unrepresented; post-settlement: reference written compensation-offer rules if discussing buyer reach; no Fair Housing flag language."

**Compliance flag — HIGH.** TCPA on unconsented FSBO SMS runs $500-$1,500 per message; FCC 2024 robotext rules tightened consent. **Fair Housing flag — MEDIUM** as scripts are a recurring NFHA testing target.

**Sample output:** "Day 1 SMS: 'Hi [Name] — Jake at [Brokerage]. Saw your Zillow listing. Maple Heights sold-to-list is 102% YTD (MLS). Happy to send the 6 closed comps if useful, no pressure. Reply STOP to opt out.'"

**Prompt block:** "I'll paste my sphere list with last-contact dates and prior-quarter topic. One personalized check-in per contact (max 90 words). Rules: reference a real shared context from my notes — never invent history; offer a market data point or resource, not a sale ask; no Fair Housing flag phrases; soft opt-out; suggest channel (email / SMS / handwritten) based on last successful one; honor consent flags."

**Fair Housing flag — LOW.** Sphere outreach rarely triggers §3604(c). **Compliance flag:** consent + opt-out required if SMS or email.

**Sample output:** "Sara — 'Hope the new quartz is holding up. Maple Heights is at 1.4 months supply (MLS, May) — your block has appreciated ~6% YoY (Zillow). Not asking for anything — just thought you'd want the number. Reply STOP to skip these.'"

**Prompt block:** "I'll paste the expired listing's MLS history (DOM, price reductions, photo count, agent remarks). Three variants: (A) market-data led, (B) photo / staging led, (C) pricing-strategy led. Rules: be specific about what changed since expiration with sourced stats; never blame the seller; never disparage the prior agent; no Fair Housing flag phrases; TCPA / CAN-SPAM compliant if SMS or email; end each with a 15-minute call ask. Max 110 words/variant."

**Fair Housing flag — LOW.** Market and process content. **Compliance flag — MEDIUM** if SMS — confirm TCPA consent per FCC 2024 rules.

**Sample output (B):** "Your home spent 73 DOM with 6 photos. Closed comps last 60d averaged 28 photos and 11 DOM (MLS). I can put a 30-photo + video shoot plan together free. Worth 15 minutes?"

**Prompt block:** "Provide contract acceptance date, financing type (cash / conventional / FHA / VA), inspection and appraisal periods, target close, state, and contingencies. Generate a date-anchored checklist. Columns: due date, item, owner (agent / buyer / lender / title / TC), source rule. Include post-settlement buyer-rep file docs — signed agreement, compensation-disclosure, broker-policy acknowledgment. Flag state-rule items [STATE VERIFY]."

**Compliance flag — HIGH.** Buyer-rep file gaps are recurring broker-audit findings post-settlement. The checklist is the lowest-cost insurance.

**Sample output:** "Day 0: signed buyer-rep on file. Day 0: compensation-disclosure form. Day 0: written compensation offer from listing broker captured. Day 1: escrow package to title. [STATE VERIFY] Day 3: lead-based paint disclosure if pre-1978."

**Prompt block:** "Brief a guest agent covering my open house. I'll paste the MLS sheet, three sold comps, and the seller's authorized talking points. Output: 5-bullet property summary (MLS sheet only); 3-bullet neighborhood context with cited stats; authorized talking points verbatim; hard-stop list — exact questions to deflect to me (price flexibility, seller motivation, outside authorization, any Fair Housing flag answer); post-settlement: how to handle a buyer arriving without a signed buyer-rep wanting a tour beyond the open house; safety check-in cadence."

**Fair Housing flag — HIGH.** Open houses are a recurring NFHA testing scenario, especially questions about "who lives in the neighborhood". The hard-stop list is the protection.

**Sample output:** "'Is this a good neighborhood for kids?' → 'I can send you school ratings after we connect — what matters most to you in a school?'. 'Will the seller take less?' → 'All offers will be presented; let's get you to a number that reflects your interest'."

**Prompt block:** "I'll paste a property highlight + listing link. Three captions (Instagram, Facebook, LinkedIn) under 220 chars each. Constraints: no Fair Housing flag language; include brokerage name + license number; 'Equal Housing Opportunity' on any image ad; no income, race, family-status, or religion targeting; every quantitative claim sourced; CTA respects post-settlement buyer-rep (offer a conversation, not a tour)."

**Fair Housing flag — HIGH.** Per HUD advertising guidance, display ads must carry the Equal Housing line. **Platform flag:** Facebook's Special Ad Category for Housing restricts targeting — review Meta's policy before boosting.

**Sample output (IG):** "Just listed — 3 bed / 2 bath ranch in Maple Heights. Updated kitchen (2023), new roof (2024). $479,000. DM to set up a conversation. [Brokerage] · License #00000 · Equal Housing Opportunity."

**Prompt block:** "I'll paste my lead-source spreadsheet: source, leads, appointments, contracts, closings, GCI, ad spend, quarter. Output: CPL, CPA, CPC by source; ROI = (GCI – spend) / spend; 90-day trend; two to scale, two to cut, with rationale; one experiment for next quarter. No invented numbers. Blank cells → [DATA NEEDED]."

**Fair Housing flag — LOW.** Internal ROI work. **Practice flag:** if a source surfaced leads via demographic-influenced placement, do not let Claude reverse-engineer demographic ROI — §3604(c) exposure.

**Sample output:** "Zillow Premier: CPL $48, CPA $290, CPC $1,840, ROI 2.4x — scale. FB lead form: CPL $14, CPA $510, CPC $4,200, ROI 0.6x — cut or restructure under Special Ad Category."

Claude as copywriter without guardrails: ships Fair Housing flag phrases into MLS public remarks, leaves the post-settlement buyer-rep paragraph stale, and generates TCPA-risk SMS. Every reported real-estate AI marketing complaint in the NFHA 2024 trends report shares this pattern.
Claude as drafting + triage assistant: scope-disciplined prompts + source-grounded stats + broker review on every commission, buyer-rep, and disclosure paragraph. Per the NAR Member Profile, the median agent runs 12 transactions a year — this workflow recovers a day per file without the liability.