Sheet & Flow Landlord Math MCP server

Sheet & Flow Landlord Math (MCP server)

<!-- mcp-name: io.github.onyxaholguin-cyber/sheet-and-flow-mcp -->

Landlord and rental-property calculators as Model Context Protocol tools, so Claude, Cursor, VS Code or any MCP client does the math the same way every time instead of estimating it.

• Offline and deterministic. No network calls, no API keys, no data leaves your computer.
• No dependencies. Plain Node.js (18+), about 380 lines you can read.
• Tested. 552 randomized cases against an independent Python model, and the worked examples from the Sheet & Flow tutorials, on Node 18, 20 and 22 in CI. The rental math was also cross-checked against spreadsheet-recalculated (LibreOffice) deals before release.

Tools

Percentages are fractions (0.25 = 25%). Dates are YYYY-MM-DD. A percentage passed as 7 instead of 0.07 is rejected with a clear message rather than silently computed.

Install

Claude Desktop (one click): download sheet-and-flow-mcp-.mcpb from the latest release and open it.

Any MCP client (stdio): clone the repo and point your client at server/index.js:

{
  "mcpServers": {
    "sheet-and-flow": { "command": "node", "args": ["/path/to/sheet-and-flow-mcp/server/index.js"] }
  }
}

It is also listed in the official MCP Registry as io.github.onyxaholguin-cyber/sheet-and-flow-mcp.

Example

Ask: "Airbnb deal: $350,000, 25% down at 7%, $185 a night, 62% occupancy, 14 blocked nights, 3.2-night stays, $95 cleaning fee, $85 cleaning cost, 15.5% host fee, 9% maintenance, $920/month fixed costs, $10,500 closing, $25,000 furnishing. Does it cash flow?"

shorttermrental_analysis returns, among other fields:

{ "monthlyCashFlow": -208.64, "keptPerBookedNight": 135.53, "breakEvenOccupancy": 0.6726, "breakEvenNightlyRate": 200.24 }

So the deal loses about $209 a month as quoted; it needs 67.3% occupancy at $185, or $200.24 a night at 62%. The same example is worked step by step in the short-term rental calculator tutorial.

How the math works

Every formula is explained in a free tutorial:

Shortened. The full README is on GitHub.