Ed Neuhaus 
Windsurf crossed a million active users in 2026, and Cognition paid $250 million to acquire it. That makes it one of the three code editors serious developers actually use right now (alongside Cursor and VS Code). So when we built the Austin MLS MCP connector, Windsurf compatibility was on the list from day one. One config block, about 90 seconds of setup, and you have live Austin property data flowing into Cascade.
If you write code and you care about Austin real estate, this is for you. And if you just heard about Windsurf and you are wondering whether it plays nice with external data sources, yeah it does. Lets walk through it.
What Windsurf Actually Is (Quick Context)
For the real estate people reading this who havent written a line of code, Windsurf is an AI-native code editor. Think VS Code but with an AI agent called Cascade built into every part of the experience. You can write code, ask Cascade questions about your codebase, generate functions from plain English, and (this is the part that matters here) connect to external data sources through MCP.
I wrote a full explainer on what MCP means for real estate if you want the background. Short version: MCP is a standard protocol that lets AI tools pull data from external sources. Instead of copy-pasting property data from a website into your editor, the editor just reaches out and grabs it.
Windsurf started life as a product from Codeium, then Cognition AI (the folks behind Devin, that “AI software engineer” everyone was talking about last year) acquired it for $250 million. So it has serious engineering muscle behind it now. And the MCP support is solid. Windsurf handles stdio, Streamable HTTP, and SSE transports, which means our remote HTTP server works out of the box.
Why You Would Want Austin MLS Data Inside Windsurf
Ok so why would anyone pipe real estate listings into a code editor? Fair question right. Lets get specific.
Developers building property tools. If you are building a property search app, an investment calculator, or anything that touches listing data, you need to test against real data constantly. Having the Austin MLS one prompt away inside Cascade means you can write a function, test it against live listings, and iterate without opening a browser. No Postman tabs. No API console. Just ask Cascade in natural language and the data comes back.
Investors who write code. I know more RE investors who write Python than most people would expect. Pull comps for a zip code, calculate cash-on-cash return, filter by days on market, screen for deals in 78738. If that sounds like you, Windsurf with MLS access turns your editor into a deal-screening machine. Ask for every listing under $600,000 in Bee Cave with 4+ bedrooms and get actual results right there in your workspace.
Agents building their own tools. This one is close to home. At Neuhaus Realty Group I have been building my own tools for years because the off-the-shelf stuff never does what I need. Windsurf with MLS access means you can prototype against live data as you build. And honestly the Cascade agent is pretty good at writing boilerplate code, which is most of what tool-building is anyway (at least for the non-developers among us, myself included).
Seth Godin’s whole point about permission marketing is that the most valuable relationship is when someone invites you in. That is what MCP does. Instead of pushing data at people through a traditional search page, we are making it available where they already work. The developer invites the data into their environment. Totally different dynamic.
Step-by-Step: Connecting Windsurf to Austin MLS Data
Here is the actual setup. I’m walking through this like you are sitting next to me.
Step 1: Open Your MCP Config File
Windsurf stores MCP server configurations in a file called mcp_config.json. The location:
- macOS:
~/.codeium/windsurf/mcp_config.json - Windows:
%USERPROFILE%\.codeium\windsurf\mcp_config.json - Linux:
~/.codeium/windsurf/mcp_config.json
If the file doesnt exist yet, create it. If it already has other servers configured, you will just add to the existing mcpServers object.
You can also get there through the GUI: click the MCPs icon in the top-right of the Cascade panel, or go to Windsurf Settings > Cascade > MCP Servers. But honestly the config file is faster if you are comfortable with JSON (and if you are using Windsurf, you probably are).
Step 2: Add the NeuhausRE MCP Server
Add this config block to your mcp_config.json:
{
"mcpServers": {
"austin-mls": {
"serverUrl": "https://mls.neuhausre.com/mcp"
}
}
}Notice the field is serverUrl here, not url. Windsurf uses serverUrl for remote HTTP servers. If you have used Cursor’s MCP setup before, that is the one difference to watch for. Cursor uses url, Windsurf uses serverUrl. No big deal right, just different syntax.
If you already have other MCP servers configured, add the "austin-mls" entry inside your existing mcpServers block. Do not replace anything.
Step 3: Restart Windsurf and Authorize
After saving the config, restart Windsurf (or reload the window). Then check Settings > Cascade > MCP Servers to verify the server loaded. You should see “austin-mls” with a green status indicator.
First time Windsurf connects, it will prompt you to authorize through OAuth. Standard OAuth 2.1 with PKCE. Windsurf handles the handshake automatically through its built-in redirect. You will see a signup form asking for your name, email, and phone number. Fill it out, authorize, done.
No API keys to manage. No npm packages to install. The server and Windsurf negotiate everything for you.
Step 4: Start Asking Cascade
Once authorized you have four tools available in Cascade:
- search_listings – Find active listings by location, price, beds, baths, features
- get_listing – Pull full details on a specific property
- get_comps – Get closed-sale comparables by address, radius, and date range
- get_market_stats – Median prices, inventory levels, days on market by location
Open Cascade and start asking in natural language. The AI figures out which tool to call. You do not need to know the tool names or pass structured arguments. Just ask like you would ask a person.
Example Prompts Once You Are Connected
So you are connected. Now what? Here are some real prompts that show what this looks like in practice. These are the kinds of questions I ask myself when I am evaluating a market or helping a client.
Searching for listings:
- “Show me active listings in Lakeway under $800,000 with at least 3 bedrooms”
- “Find homes in Bee Cave with a pool that have been on market more than 30 days”
- “What is available in 78746 between $500K and $750K?”
Pulling comps:
- “Get comparable sales within half a mile of 123 Main St, Bee Cave TX from the last 6 months”
- “Show closed sales in Dripping Springs over $700K in the last 90 days”
Market stats:
- “What is the median home price in Bee Cave right now?”
- “Show me inventory levels and days on market for Austin zip code 78738”
- “Compare median prices in Dripping Springs vs Bee Cave”
But here is where Windsurf specifically gets interesting. Because Cascade is an AI agent (not just a chat window), you can chain these queries into workflows. Ask for listings in a zip code, then immediately ask Cascade to calculate price per square foot across the results, then ask it to write a Python function that screens for deals below the median. The data and the code live in the same environment. That is the whole point.
How Windsurf Compares to Other MCP Clients
If you have been following this series, you know we have setup guides for Cursor, Claude Desktop, and Perplexity. So how does Windsurf stack up?
Windsurf vs Cursor. Both are AI code editors built on VS Code. The setup is nearly identical, just different config file locations and one field name difference (serverUrl vs url). If you already use one, switching is trivial. Windsurf’s Cascade agent is arguably more autonomous than Cursor’s Composer for multi-step tasks, but the MLS tools work the same in both.
Windsurf vs Claude Desktop. Different use cases entirely. Claude Desktop is for research and analysis. Windsurf is for building things. If you want to ask questions about the Austin market and get thoughtful answers, Claude is your tool. If you want to build an app that uses Austin listing data, Windsurf is your tool.
Windsurf vs Perplexity. Perplexity is a search engine with MCP support. Great for one-off queries and research. Windsurf is a development environment. Different audiences, different workflows. Some people use both (Perplexity for quick lookups, Windsurf for building).
The MCP protocol is the same across all of them. Once you understand how it works in one client, the concept transfers everywhere. I wrote a beginner’s guide to AI for real estate if you want the broader context on how all this fits together.
Free Access and the Active Buyer Retainer
The NeuhausRE MCP connector is free to use. You get 10 results per search, 10 queries per hour, and 100 queries per month. That is plenty for exploring, building prototypes, and casual property research.
If you are actively searching for a home in Austin, we also offer an Active Buyer retainer at $200/month. That bumps your throughput to 50 results per search, 200 queries per hour, and 10,000 per month, plus unlocks comps and market stats. Think of it as a buyer representation retainer with me (Ed Neuhaus, actual licensed broker), not a data subscription. The entire amount gets credited back to you at closing if you purchase through Neuhaus Realty Group. So if you end up buying a home in Austin, you effectively paid nothing.
The property data itself is the same listing data displayed on NeuhausRE.com. We are a VOW licensee displaying our own data through an additional channel. The retainer is about working with me as your buyer’s agent, not purchasing access to a database.
One Thing Worth Knowing About Windsurf’s Tool Limit
Windsurf caps active MCP tools at 100 across all your connected servers. Our Austin MLS server exposes 4 tools, so this is not going to be a problem unless you have a ton of other MCP servers loaded. But if you are the kind of person who connects every MCP server you find (and I respect that instinct, I do the same thing with browser extensions), just know the ceiling exists. You can toggle servers on and off in the Cascade panel, no big deal.
Why I Built This
I get asked about this so lets just address it directly. Most real estate websites make you click through page after page to find what you need. Search, filter, click, back, search again. It works but it is slow if you are doing serious analysis.
I built this connector because the same property data we display on NeuhausRE.com should be accessible in the tools people actually work in. Not just through a traditional search page (although we have that too). Not a chatbot that asks three questions and then demands your phone number before it tells you anything useful. The same listings, the same comps, the same market data, just delivered through your AI tools instead of a web browser.
And honestly it is good for my business too. People who use this data seriously tend to be serious buyers. The kind of person who pulls comps in a Python script before they ever call an agent? Those are my favorite clients. We tend to get along pretty well.
Frequently Asked Questions
Ready to Connect?
Setup is genuinely 90 seconds. Add the config block, restart Windsurf, authorize, done. And if you run into issues or want to talk about what you are building, reach out to me directly. I am always interested in what developers and investors are doing with programmatic access to our property data. Some of the coolest tools I have seen started as side projects in exactly this kind of setup. And I am not exactly a professional developer myself (lighting designer turned realtor turned… whatever I am now), so I appreciate when people show me what they built.
Be safe, be good, and be nice to people.
