AgencyGraph on MCP: Ask Claude, Cursor or Other MCP Clients Where Your Estate Agency Stands Against Every Competitor

CASAFARI's AgencyGraph over MCP lets you connect Claude, Cursor and other MCP clients to benchmark estate agency market share across about 41,000 agencies and offices with homes for sale in 4,767 Spanish towns and about 17,600 in 309 Portuguese towns (weekly build, October 2026). Querying five core metrics directly via Model Context Protocol allows agency owners, franchise networks, vendors and investors to evaluate performance and track competitor watchlists week by week.

How can you see where your agency stands against every competitor in your town?

Whether you are preparing for a valuation meeting tomorrow morning in Málaga or Cascais where a homeowner asks why they should instruct your firm, or evaluating new territories for network expansion, immediate access to structured competitor figures changes the conversation. Instead of relying on anecdotal claims or manual web browsing, you can query market share data directly within your preferred AI workspace.

CASAFARI, the property data layer for estate agents and their AI assistants, builds this estate agency database every week by indexing advertisements for homes across 4,767 Spanish and 309 Portuguese towns. The system collects property advertisements published on portals and individual agency websites, cleans the records, and deduplicates multiple advertisements into a single property entity. Every event—new instructions, price reductions, price increases, delistings, and relistings—is tracked over time. AgencyGraph resolves variations in business names into unified agency entities, attributes active stock and market events to each firm, and measures performance across every municipality where the agency operates. It focuses strictly on corporate entities, including independent agencies, branch offices, and franchise networks, without collecting or returning individual agent details.

You can test these queries immediately by starting a 30-day free trial with 1,000 tokens at platform.casafari.com/signin or entering an example prompt into Claude, Cursor or another MCP client.

How to calculate your local market share and benchmark competitors on five measures

Objective market share figures provide clear evidence of an agency's standing within a town across specific home categories, including all residential properties, apartments, and houses for sale (rental properties are not currently measured). AgencyGraph calculates five distinct metrics using asking prices from indexed advertisements, never closing prices:

1. Portfolio value: total, mean, median, minimum, maximum, and quartile asking prices in EUR, alongside the median asking price per m².
2. Growth: new instructions and withdrawn homes during the latest complete quarter, compared against the preceding quarter and the corresponding quarter of the previous year.
3. Price adjustments: total counts of price reductions and increases, the percentage of portfolio properties with a price cut, and the median percentage change for both cuts and increases.
4. Initial pricing gap: the median difference between an agency's initial asking prices per m² and the town's overall median, the proportion of instructions launched more than 10% above the local median, and how many of those overpriced homes were subsequently reduced.
5. Time on market: minimum, maximum, median, and mean days on the market.

These rankings represent a digital inventory index based on online advertised stock as detected by CASAFARI, rather than total business revenue or off-market transactions. Because a property advertised concurrently by two agencies is counted once for each firm, total agency market shares within a town can sum to more than 100%. To maintain statistical integrity, a town or agency with fewer than 5 homes shows its count without figures, returning 'not known' rather than guessing. Private sellers, financial institutions, property developers, and MLS feeds are filtered out, while franchise branches are recorded as distinct agency entities associated with their parent network. In Spain, approximately 22,000 agencies maintain 5 or more active homes for sale and receive full metric profiles; in Portugal, around 6,600 agencies meet this threshold. Italy, Germany, Austria, Poland and Greece come next, then France, Belgium and Luxembourg, each once its weekly build passes its checks.

In 150 Spanish and Portuguese towns, only about 7–13% of the sources AI assistants cite are an agency's own website; about 35% are portals (CASAFARI GEO study, October 2026, 1,199 questions).

Coming to AgencyGraph, at no extra price (no dates promised): tech stack detection (CRM and website vendor), MLS membership, specialisation (luxury, new builds, rentals, commercial, foreign buyers), listing quality, exclusivity, portal mix, SEO, AEO and GEO scores, and online presence (Google profile, ads, social profiles).

Explore every tool and metric on our capabilities page, view full sample reports in our sample report index (in English), check our main AgencyGraph product page, or review the AgencyGraph documentation.

How does the competitor watchlist track up to 25 estate agencies week by week?

Rather than performing manual portal searches every Monday morning, managers can maintain a dedicated competitor watchlist. You can select up to 25 agencies—including your own firm and key local competitors—and query changes recorded since any date within the past year.

The watchlist report details new instructions, delistings, relistings, price increases, and price reductions with their median adjustment percentages, as well as corporate merges. Watchlist requests cost 25 tokens per call for up to 25 agencies. Information is retrieved on demand whenever you run a query; AgencyGraph does not send automated push notifications over MCP or API.

Which tools does the AgencyGraph MCP server offer and what do they cost in tokens?

AgencyGraph uses transparent token pricing. Resolving agency names, looking up country coverage, and looking up location IDs incur no token charges. You are billed only when a tool returns a successful answer.

Tool Name	Function	Token Price
agencygraph_coverage	Returns supported countries and their latest weekly build dates	Free
agencygraph_country_coverage	Returns town counts and total measured agencies for a country	5 tokens per call
agencygraph_resolve_agency	Finds an agency by its name, a trading name, or website URL	Free
agencygraph_agency_profile	Full profile: trading names, websites, network, totals, new and delisted homes for four quarters, and largest towns (up to 25) with measures and rankings	50 tokens per call (€2)
agencygraph_area_benchmark	Ranks a town's agencies by a chosen measure (up to 100 agencies)	75 tokens per page (max 25) (€3)
agencygraph_market_share	Returns agency shares of active, new, and delisted homes in a town	50 tokens per page (max 25)
agencygraph_territory	Lists all towns where an agency operates, with share and portfolio value	50 tokens per page (max 25)
agencygraph_network	Network summary and full list of affiliated branch offices	75 tokens per page (max 25)
agencygraph_screen_agencies	Filters agencies by ranges of measures, size, network status, or property type	100 tokens per page (max 25) (€4)
agencygraph_changes	Watchlist: weekly changes (new, delisted, price changes, merges) for up to 25 agencies	25 tokens per call
REST City Dataset	Complete download of a town's agencies, metrics, and four-quarter history	Small (<400): 2,500 tokens (€100) Mid (400–1,999): 6,250 tokens (€250) Large (2,000+): 12,500 tokens (€500)

City datasets can be downloaded again for a year, rebuilt from the newest weekly build. Custom orders are available for full country extracts. Note that AgentGraph AI for agencies and individual agents in Germany remains a separate product with its own toolsets and token pricing.

How do you connect Claude, Cursor or other MCP clients to AgencyGraph?

Connecting AI assistants to AgencyGraph provides live data across 41,000 Spanish and 17,600 Portuguese agencies.

1. Copy the MCP server URL: https://platform.casafari.com/mcp/agencygraph.
2. In Claude: go to Settings → Connectors → Add custom connector, paste https://platform.casafari.com/mcp/agencygraph, and sign in. In Cursor or other MCP clients: enter the same address and complete the OAuth sign-in in your browser.
3. No manual API keys required.

For custom integrations, developers can access 15 REST API endpoints under /v1/agencygraph/{country}/..., documented on developers.casafari.com. You can review our dedicated AgencyGraph MCP page to get started.

What can you ask Claude, Cursor or your AI assistant about estate agencies?

When asking AI assistants about local market dynamics, structured queries yield clear context. Here are five example prompts for common use cases:

• Agency owner (where I stand):

  "Rank the real estate agencies in Málaga by homes for sale, with their price cuts and median days on market, and show where my agency sits." Tools used: agencygraph_area_benchmark (75 tokens a page of 25), agencygraph_agency_profile (50 tokens). See the Málaga agency market share sample report (in English).

• Agency owner (win the mandate):

  "Compare my agency's overpricing at start and time on market in Cascais with the five agencies that hold the most homes there." Tools used: agencygraph_agency_profile (50 tokens), agencygraph_market_share (50 tokens a page of 25). See the Cascais winning instructions sample report (in English).

• Franchise network (recruitment and territory gaps):

  "Find independent agencies in Valencia province with 30 to 150 homes for sale that grew this quarter and cut prices less than the town's median." Tools used: agencygraph_screen_agencies (100 tokens a page of 25), agencygraph_territory (50 tokens). See the network recruitment sample report and network territory gaps report (both in English).

• Vendors selling to agencies:

  "List agencies in Porto with more than 50 homes for sale whose homes stay on the market longer than the town's median." Tools used: agencygraph_screen_agencies (100 tokens a page of 25), REST city dataset on the API (2,500–12,500 tokens a year). See the vendor prospect list sample report (in English).

• Banks, developers and investors:

  "Which agencies in Marbella move houses off the market fastest, with the least overpricing at start, among those with at least 40 homes for sale?" Tools used: agencygraph_area_benchmark (75 tokens), agencygraph_screen_agencies (100 tokens). See the investor agency selection sample report (in English).

For custom workflows and code examples, explore our AgencyGraph MCP recipes for AI agent builders. You can also read detailed guides for estate agency competitor analysis, franchise network expansion, property portal market coverage, agency evaluation for developers and investors, and submitting data updates on our AgencyGraph corrections page. To begin analysing your market, claim your 1,000-token free trial at platform.casafari.com/signin.