À propos
Cette compétence apporte une connaissance du domaine de l'immobilier résidentiel pour éclairer les spécifications techniques, évitant ainsi des hypothèses CRUD simplistes. Elle formalise des concepts clés tels que les données MLS/IDX, les cycles de vie des annonces et la coordination des transactions pour garantir que l'architecture reflète la réalité du secteur. Utilisez-la lors de la spécification de produits proptech comme les systèmes d'annonces, les CRM ou les outils de transaction, afin d'établir des modèles de domaine précis.
Installation rapide
Claude Code
Recommandénpx skills add avelikiy/great_cto -a claude-code/plugin add https://github.com/avelikiy/great_ctogit clone https://github.com/avelikiy/great_cto.git ~/.claude/skills/vertical-real-estateCopiez et collez cette commande dans Claude Code pour installer cette compétence
Documentation
Vertical: residential real estate — don't spec it naively
Real estate looks like generic CRUD (a listing is a record, a lead is a contact, a deal is a checklist). It isn't. The domain has hard-won structure — MLS quirks, status lifecycles, months-long sales cycles, agent-vs-brokerage data boundaries — that a naive build ignores and then rebuilds. This skill front-loads that structure so the spec is right the first time.
Incumbents to know (and what they own): Lone Wolf / Propertybase (brokerage CRM + back office), Follow Up Boss (lead-to-close CRM, the nurture gold standard), CINC (lead-gen
- CRM), Top Producer (legacy CRM), kvCORE / BoldTrail (all-in-one platform). They are expensive, broad, and switching-cost-heavy — which is why the wedge matters (see per-product).
1. Domain vocabulary (use these terms in the spec)
- MLS (Multiple Listing Service) — regional database of listings; there are ~500+ MLSs in the US, each its own system, login, and field set. There is no single national MLS.
- IDX (Internet Data Exchange) — the rules + feed that let a brokerage display other brokers' MLS listings on its own site. Governed by per-MLS display/redistribution rules.
- RESO — the standards body. RESO Web API (modern REST/OData feed) and RESO Data Dictionary (canonical field names) are the standard — but adoption and cleanliness vary per MLS. "RESO-compliant" still means per-MLS quirks.
- Listing status — lifecycle, not a flag:
active → pending / contingent → closed(pluscoming soon,active under contract,withdrawn,expired,sold). Status drives display rules and downstream automation. - Buyer agent vs seller (listing) agent — opposite sides of a transaction; data and permissions differ. A contact can be a buyer lead and later a seller.
- Brokerage vs agent — the brokerage holds the license and (often) the data; the agent is the user. Commission split is how the deal's commission divides brokerage↔agent.
- Escrow / settlement / closing — the funded close; contingencies (inspection, financing, appraisal) are conditions that must clear first, each with a deadline.
- Transaction coordinator (TC) — the person who shepherds a deal from accepted-offer to close: chasing docs, signatures, and deadlines. Often done in Dotloop / Excel today.
- CMA (Comparative Market Analysis) — comp-based price estimate an agent gives a seller.
- Lead-to-close funnel — capture → nurture → active → under-contract → closed; months long.
- Syndication portals — Zillow, Realtor.com, Redfin, Trulia, etc.; a listing is pushed (syndicated) to many; our copy is one of many downstream copies.
2. Non-obvious domain rules (the stuff that breaks naive specs)
- There is no one MLS schema. Each MLS has its own auth, field quirks, photo handling, and redistribution rules. RESO standardizes the intent but the data is messy per-MLS. Model an integration adapter per MLS, not one global importer.
- Transaction-coordination is the underserved, high-pain wedge. It's the part still done in Dotloop/Excel with manual deadline-chasing. Low switching cost (it's not the system of record for leads), high pain, clear ROI. Lead this if choosing where to land first.
- A listing has a canonical source and many syndicated copies. The MLS record (or our
record) is the source of truth; portal copies derive from it. Don't model portal copies as
independent listings — model
source+syndication targets. - Lead nurture is long-cycle. A real-estate lead can sit warm for 6–18 months. The CRM's job is not close-this-week; it's stay-top-of-mind for months with automated drip until the lead is ready. A short-funnel CRM design is simply wrong for this domain.
3. What a naive build gets wrong
- ❌ One MLS schema. Assuming a single import format. Reality: per-MLS adapter, per-MLS redistribution rules, RESO Data Dictionary as the target normalization, not the source.
- ❌ Listing without status lifecycle + syndication canonical. A flat "listing" row with no status state machine and no source/target model can't drive display rules or feed portals.
- ❌ TC checklist without deadline + contingency tracking. A plain task list misses the point: the value is enforced deadlines and contingency clearing, with alerts.
- ❌ Lead CRM without long-cycle nurture. Stage + next-touch + multi-month drip is the core; a pipeline-only CRM (close/lost in weeks) doesn't fit.
- ❌ Ignoring agent-vs-brokerage data boundaries. Who owns the lead and listing data — agent or brokerage — is a real permission/ownership boundary. Bake it into the model, not bolt it on later.
4. Must-model entities
Seed the domain model with these (exact fields negotiable; the shape is not):
- Listing —
status(lifecycle state machine, §1),source_ref(MLS id / RESO key — the canonical source),syndication_targets[](Zillow/Realtor.com/… with per-target state), price, address, beds/baths, floorSize, photos, listing_agent, brokerage. - Lead —
stage(capture→nurture→active→under-contract→closed),last_contact_at, owning agent, source, buyer/seller intent, and an attached long-cycle nurture (drip campaign, next-touch date). Tie to [[lifecycle-messaging]]. - Transaction —
checklist[](tasks),deadlines[](each dated + owner + alert),contingencies[](inspection/financing/appraisal, each with clear-by date + status),documents[](e-signed), parties (buyer/seller/agents/TC), close date. - Property / Unit + MaintenanceRequest (status, priority, tenant, assignee, photos), Lease/rent (amount, due date, tenant), tenant-comms thread. (property-mgmt.)
5. Per-product notes (wedge + the one domain thing to get right)
- listings (content) — the one thing: MLS/IDX correctness — per-MLS adapter,
source_refcanonical, status lifecycle, redistribution-compliant display. Pairs with [[local-seo]] (listings must rank + syndicate; our page is the canonical, portal copies point back) and [[migration-ready-schema]] (source_ref= MLS/RESO key for re-import/dedupe). - lead-crm (crm) — the one thing: long-cycle nurture — months-long automated drip, stage + last-contact, top-of-mind not close-now. Competes with Follow Up Boss; the bar is nurture quality. Pairs with [[lifecycle-messaging]].
- transaction-coordination (crud) — the one thing: this IS the wedge — low switching cost, high pain, replaces Dotloop/Excel. Get deadline + contingency tracking right (dated, owned, alerted); the checklist alone is table stakes. Recommend landing here first.
- property-mgmt (crud) — the one thing: maintenance-request + rent + tenant-comms as first-class flows; don't reduce it to a generic ticket list. Don't hold tenant/owner funds naively (see §6 escrow/trust).
6. Compliance (light — flag for the reviewer, don't solve here)
- MLS / IDX redistribution + display rules — per-MLS; controls what may be displayed, for how long, with what attribution. The integration adapter must honor them.
- Fair Housing Act — listing/ad copy and targeting must not discriminate (protected classes). Applies to listing descriptions and any lead-targeting automation.
- Escrow / trust-account basics — do not hold or move client/tenant funds naively. Earnest money and rent flow through trust/escrow with strict accounting; integrate a compliant provider rather than building a wallet.
- E-sign — transaction docs need legally-valid e-signature (ESIGN/UETA); use a real e-sign provider, capture audit trail.
Output
When applied, contribute a Domain model block to the architecture/design doc:
## Domain model (real estate)
- entities: Listing(status lifecycle + source_ref + syndication_targets) · Lead(stage + last_contact + nurture) · Transaction(checklist + deadlines + contingencies + docs) · Property/Unit + MaintenanceRequest
- wedge: <which product lands first + why> (default: transaction-coordination)
- MLS/IDX: per-MLS adapter · RESO Data Dictionary as normalization target · redistribution rules honored
- nurture: long-cycle (6–18mo) drip, not short funnel
- compliance flags: IDX display rules · Fair Housing (copy) · escrow/trust (no naive funds) · e-sign
Dépôt GitHub
Frequently asked questions
What is the vertical-real-estate skill?
vertical-real-estate is a Claude Skill by avelikiy. Skills package instructions and resources that Claude loads on demand, so Claude can perform vertical-real-estate-related tasks without extra prompting.
How do I install vertical-real-estate?
Use the install commands on this page: add vertical-real-estate to Claude Code as a plugin, or clone its repository into your skills directory, then restart Claude so it picks up the skill.
What category does vertical-real-estate belong to?
vertical-real-estate is in the Other category, tagged ai.
Is vertical-real-estate free to use?
Yes. vertical-real-estate is listed on AIMCP and free to install. It runs inside Claude, so no separate service account is required to use the skill itself.
Compétences associées
LlamaGuard est le modèle de Meta, doté de 7 à 8 milliards de paramètres, conçu pour modérer les entrées et sorties des LLM selon six catégories de sécurité comme la violence et les discours haineux. Il offre une précision de 94 à 95 % et peut être déployé avec vLLM, Hugging Face ou Amazon SageMaker. Utilisez cette compétence pour intégrer facilement le filtrage de contenu et des garde-fous de sécurité dans vos applications d'IA.
Cette compétence de Claude aide les développeurs à optimiser les coûts du cloud grâce au redimensionnement des ressources, aux stratégies d'étiquetage et à l'analyse des dépenses. Elle fournit un cadre pour réduire les dépenses cloud et mettre en œuvre une gouvernance des coûts sur AWS, Azure et GCP. Utilisez-la lorsque vous devez analyser les coûts d'infrastructure, redimensionner les ressources ou respecter des contraintes budgétaires.
Cette compétence quantifie les LLMs en précision 8 bits ou 4 bits à l'aide de bitsandbytes, permettant une réduction de 50 à 75 % de la mémoire utilisée avec une perte de précision minime. Elle est idéale pour exécuter des modèles plus volumineux sur une mémoire GPU limitée ou pour accélérer l'inférence, prenant en charge des formats comme INT8, NF4 et FP4. La compétence s'intègre à HuggingFace Transformers et permet l'entraînement QLoRA ainsi que l'utilisation d'optimiseurs en 8 bits.
Cette compétence Claude analyse les marchés des paris sportifs, incluant les spreads, les over/under et les paris spéciaux, en examinant les tendances historiques et les statistiques situationnelles pour identifier les paris à valeur ajoutée. Elle fournit une sortie en markdown structuré avec des recommandations actionnables à des fins éducatives. Les développeurs doivent l'utiliser pour des outils d'analyse de paris sportifs tout en notant qu'elle est conçue uniquement pour le divertissement et l'éducation.
