DSCR Loan Calculator MCP

DSCR Loan Calculator MCP

Exact DSCR loan math for AI assistants: debt service coverage ratio, max loan, rent needed, scenario comparison and amortization for rental property loans.

Hosted MCP URL (no sign-in, no token): https://dscr-loan-calculator-mcp.lipmichal.workers.dev/mcp

Money site with the full calculators and rate pages: dscrradar.com

Tools

Every result has structured JSON, a one-paragraph summary, an as_of date, a source URL and a more link to the matching dscrradar.com page.

What it knows that your model does not

• Cent-exact payments. Language models often get the annuity formula wrong by a few dollars, or round in the wrong place. This server uses the standard annuity formula with monthly compounding and rounds each part half up to the cent, the way a lender schedule does.
• The real definition of PITIA (Fannie Mae Selling Guide B3-6-03): principal, interest, taxes, insurance, association dues, ground rent, special assessments.
• Exact tier edges. A DSCR of 1.2499 is not 1.25. Tiers are decided on the exact ratio with integer math, and the 2-decimal DSCR is truncated, not rounded up.
• Solvers. Max loan and rent needed are solved exactly against the rounded payment, not estimated.
• How lenders actually differ. Some lenders look for 1.25, some accept 1.00, some go down to 0.75, and some offer no-ratio programs. Sources are in each result.

Install

Claude.ai custom connector, ChatGPT developer mode

Paste this URL as a custom connector. No token or sign-in.

https://dscr-loan-calculator-mcp.lipmichal.workers.dev/mcp

Claude Code

claude mcp add --transport http dscr-loan-calculator https://dscr-loan-calculator-mcp.lipmichal.workers.dev/mcp

Or run it locally over stdio:

claude mcp add dscr-loan-calculator -- npx -y github:theluckystrike/dscr-loan-calculator-mcp

Claude Desktop

Add to claude_desktop_config.json:

{
  "mcpServers": {
    "dscr-loan-calculator": {
      "command": "npx",
      "args": ["-y", "github:theluckystrike/dscr-loan-calculator-mcp"]
    }
  }
}

Cursor

Add to ~/.cursor/mcp.json:

{
  "mcpServers": {
    "dscr-loan-calculator": { "url": "https://dscr-loan-calculator-mcp.lipmichal.workers.dev/mcp" }
  }
}

VS Code

Add to .vscode/mcp.json:

{
  "servers": {
    "dscr-loan-calculator": { "type": "http", "url": "https://dscr-loan-calculator-mcp.lipmichal.workers.dev/mcp" }
  }
}

Test with curl

curl -s -X POST https://dscr-loan-calculator-mcp.lipmichal.workers.dev/mcp \
  -H 'content-type: application/json' -H 'accept: application/json, text/event-stream' \
  -d '{"jsonrpc":"2.0","id":1,"method":"tools/call","params":{"name":"calculate_dscr","arguments":{"monthly_rent":2500,"loan_amount":300000,"annual_rate_pct":7,"annual_property_tax":3600,"annual_insurance":1200}}}'

Example prompts

• "Rent is $2,500, the loan is $300,000 at 7% for 30 years, taxes $3,600 a year and insurance $1,200. What is my DSCR?"
• "What is the biggest DSCR loan I can get on a $400,000 rental that rents for $3,000, at 7.5%, if the lender wants 1.25?"
• "How much rent do I need for a 1.25 DSCR on a $300,000 loan at 7%?"
• "Compare 20% down at 7%, 25% down at 7.5% and 30% down interest-only at 8% on a $400,000 property renting for $3,200."
• "Show the balance after 10 and 20 years on a $500,000 loan at 7.5%, 10 years interest-only then 20 years amortizing."

Math and rounding

• Monthly rate = annual rate / 12. P&I = L x r / (1 - (1 + r)^-n). At 0%, P&I = L / n.
• Interest-only payment = L x r.
• Monthly tax and insurance = annual amount / 12.
• Each of these is rounded half up to the cent before they are added into PITIA.
• Qualifying rent = gross rent x (1 - vacancy), rounded half up to the cent. Vacancy defaults to 0, because many DSCR lenders use gross rent.
• DSCR = qualifying rent / PITIA. Shown truncated to 2 decimals and as an exact ratio to 6 decimals.
• Amortization runs month by month. Interest is rounded half up to the cent each month and the final payment clears the balance.

23 test fixtures check these numbers. The expected values come from a separate Python Decimal script (scripts/expected.py) that uses brute-force search, not from the server code.

Data sources

See DATA_SOURCES.md. In short, the PITIA definition is from the Fannie Mae Selling Guide, and the lender bands are from Griffin Funding, Defy Mortgage and LendingOne pages fetched on 2026-10-30.

Limitations

• This is math, not a loan offer or approval. Lender minimums, rate pricing, reserve rules and LTV caps vary by lender and change often.
• DSCR here is the residential DSCR-loan measure (rent / PITIA). Commercial lenders often use NOI / annual debt service instead. Some lenders subtract vacancy or expenses first; use vacancy_pct and monthly_other_housing to model that.
• Some lenders qualify interest-only loans on the amortizing payment. calculate_dscr shows both when interest_only is true.
• The server does not quote live rates. Current DSCR rate pages are on dscrradar.com.

Development

npm install
npm run build
npm test
npx wrangler deploy

src/core.ts holds the pure math. src/stdio.ts and src/worker.ts are thin entries over the same tool registry in src/tools.ts.

License

MIT

Part of the Zovo MCP collection: https://mcp.zovo.one

More data servers from the same author

Each one is free, needs no API key and has a hosted URL you can paste into Claude or ChatGPT.

The paid back-office collection (invoices, VAT, PDFs, spreadsheets) is at https://mcp.zovo.one