Skip to main content

​
Connect with AI agents

Prop AI Deals is AI-native by design. Every major AI agent ecosystem can discover and use our API documentation without a single line of glue code on your side. Three integration layers, in order of richness:
LayerBest forURL
MCP ServerClaude Code, Cursor, Claude Desktop, Continue, Cline, any MCP clienthttps://docs.propaideals.co.uk/mcp
llms.txtChatGPT browsing, Perplexity, Claude search, You.comhttps://docs.propaideals.co.uk/llms.txt
llms-full.txtOne-shot context loading for any LLMhttps://docs.propaideals.co.uk/llms-full.txt
Our docs are exposed as a Model Context Protocol server with two tools:
  • search_prop_ai_deals — semantic + keyword search across all our API docs
  • query_docs_filesystem_prop_ai_deals — read individual doc pages directly
When connected, your AI agent can answer questions about our API, generate working code samples for any endpoint, and look up plan details, rate limits, and error codes — all without the user pasting docs into the chat.

​
Claude Code

Add the MCP server with one CLI command: 
claude mcp add propaideals https://docs.propaideals.co.uk/mcp
Then in any Claude Code session you can ask: 
> How do I search for properties in London with a yield above 8%?
> What error code do I get when I exceed the monthly cap?
> Generate a Python script that fetches sold history for a property
Claude Code will use the MCP server to look up real answers from our docs.

​
Cursor

Add to your ~/.cursor/mcp.json: 
{
  "mcpServers": {
    "propaideals": {
      "url": "https://docs.propaideals.co.uk/mcp"
    }
  }
}
Restart Cursor. The Cursor chat will now have a “propaideals” tool available — it’ll auto-invoke when you ask property API questions.

​
Claude Desktop

Edit your claude_desktop_config.json (located in ~/Library/Application Support/Claude/ on macOS, %APPDATA%\Claude\ on Windows): 
{
  "mcpServers": {
    "propaideals": {
      "type": "http",
      "url": "https://docs.propaideals.co.uk/mcp"
    }
  }
}
Restart Claude Desktop. You’ll see a hammer icon (🔨) in the chat input — click it to confirm the propaideals server is connected.

​
Continue (VS Code / JetBrains)

Edit ~/.continue/config.json: 
{
  "experimental": {
    "modelContextProtocolServers": [
      {
        "transport": {
          "type": "http",
          "url": "https://docs.propaideals.co.uk/mcp"
        }
      }
    ]
  }
}

​
Cline (VS Code)

In Cline settings → MCP Servers → Add Server → enter https://docs.propaideals.co.uk/mcp.

​
Other MCP clients

Any MCP-compatible client should work. The transport is HTTP streaming (not stdio), and authentication is public read-only — no API key needed to read the docs.

​
llms.txt (for non-MCP agents)

If your agent doesn’t speak MCP yet, it almost certainly knows about llmstxt.org. Our llms.txt is a structured Markdown index of all the API docs: 
https://docs.propaideals.co.uk/llms.txt
https://www.propaideals.co.uk/llms.txt
Both URLs serve the same file. ChatGPT browsing, Perplexity, Claude search, You.com, and most “research mode” agents will discover and use it automatically when answering questions about our API. You can also paste it into any chat: 
Here's the Prop AI Deals API documentation index:
https://docs.propaideals.co.uk/llms.txt

Help me build a Python script that searches for high-yield properties.

​
llms-full.txt (one-shot context)

For agents that prefer to load all docs into context at once (instead of fetching pages incrementally), we provide a 2,400+ line concatenated file: 
https://docs.propaideals.co.uk/llms-full.txt
https://www.propaideals.co.uk/llms-full.txt
This is the entire public API documentation — quickstart, authentication, rate limits, errors, response format, every endpoint reference, and the changelog — in one Markdown file. Useful for:
  • Long-context models that can ingest the whole thing in a single message
  • Pre-loading the docs into a system prompt
  • Building a custom RAG index of our docs offline
  • Agents that don’t want to make N HTTP calls during a conversation

​
What this means for your apps

If you’re building a property tool and want users to be able to ask natural-language questions like:
“Find me 3-bed houses under £400k with at least 7% yield in Manchester”
…you have three options:
  1. Use our AI chat endpoint directly — we run the LLM, you get back property cards. Costs 5× a normal request.
  2. Connect your AI agent to our MCP server so it can look up the right endpoint to call, then call it itself with your paid_* key. The agent does the orchestration; you pay our standard rate.
  3. Embed our llms.txt in your system prompt so any LLM you use already knows our API surface.
Option 1 is fastest to ship. Option 2 gives you the most control. Option 3 is the cheapest.

​
Verifying the MCP server

You can test the MCP server is reachable without an MCP client by sending a JSON-RPC call directly: 
curl -X POST https://docs.propaideals.co.uk/mcp \
  -H "Content-Type: application/json" \
  -H "Accept: application/json, text/event-stream" \
  -d '{"jsonrpc":"2.0","method":"tools/list","id":1}'
You should get back a list of two tools: search_prop_ai_deals and query_docs_filesystem_prop_ai_deals.

​
Status & support

The MCP server is hosted by Mintlify and inherits their uptime. We monitor it as part of our status page. If you have an issue connecting an AI agent or want a feature, email api@propaideals.co.uk.