Developers & AI agents

Give your AI a real underwriting engine

REIzer's MCP server connects Claude, Cursor, ChatGPT β€” any MCP-compatible client β€” to the same engine that powers the REIzer analyzers: all six investment strategies, sale & rental comps, and STR market data. Your AI stops doing head-math and starts calling real tools.

  • 6 purpose-built tools
  • Free core analysis β€” no key
  • One-command setup
Terminal
$ claude mcp add --transport http reizer https://reizer.io/mcp
βœ“ connected β€” 6 tools available
6strategies covered
0–100deterministic deal score
7–12stress scenarios per deal
30-dayshareable reports
Docs navigation

From setup to analysis in 30 seconds

1

Add the server

One command or one config entry β€” per-client instructions. No SDK, no code, no account.

2

Ask in plain English

"Is this duplex a good deal at $285K?" Your AI picks the right tool and calls the engine.

3

Get real numbers + a report

Score, cash flow, stress tests β€” plus a shareable report page with every analysis.

See it in action

A real session β€” these are the engine's actual outputs, not mock numbers. Your AI chains the tools on its own:

claude β€” reizer-mcp-session
> I found a duplex listed at $285K that should rent for $3,100/mo.
  25% down at 6.75% β€” good deal?

Called analyze_deal(strategy: "ltr", price: 285000, marketRent: 3100, …)

    REIzer Deal Score

    Score:             96 / 100 β€” Strong
    Monthly cash flow: $637
    DSCR:              1.46      Cap rate:       8.5%
    Cash-on-cash:      9.9%      Cash invested:  $77,250
    Stress tests:      survives all 7 (incl. recession combo)

Strong deal β€” it clears every stress scenario with room to spare.
The full shareable report: https://reizer.io/report.html?id=…

> What's the most I could offer and still clear $400/mo
  at 8% cash-on-cash?

Called solve_max_offer(strategy: "ltr", targets: {monthlyCF: 400, cocReturn: 0.08}, …)

    Max offer:         $303,657
    Binding target:    cash-on-cash β‰₯ 8%  (not the cash flow)
    At that price:     $546/mo Β· DSCR 1.37

Anything under list is comfortably inside your targets β€” you have
$18K of negotiating headroom above asking before the deal thins out.

Works with your favorite AI tools

MCP (Model Context Protocol) is the open standard for connecting AI models to external tools β€” adopted across the major AI clients. REIzer's server speaks standard streamable HTTP, so anything below connects with a URL:

Claude Codeone terminal command
Claude Desktop & claude.aicustom connector
ChatGPTMCP connector
Cursorone mcp.json entry
VS CodeCopilot agent mode
Agent frameworksany streamable-HTTP host

Six tools, the full REIzer surface

Analysis tools

4 tools Β· free

The complete underwriting engine β€” six strategies, deterministic scoring, stress testing, and offer solving. No key, no account.

analyze_deal

Full underwrite of one strategy: 0–100 score, metrics, stress tests, thresholds, and a shareable report link.

best_use

Runs every strategy the property qualifies for and ranks them β€” "should this be a flip, a BRRRR, or a rental?"

solve_max_offer

Solves the maximum purchase price that still hits your targets β€” cash flow, DSCR, cash-on-cash, or flip margin.

list_strategies

Machine-readable input schema for each strategy β€” what to send, what's required, what defaults exist.

Licensed data tools

2 tools Β· API key

The same licensed data feeds the REIzer app uses, metered against your plan's monthly allowances (Investor and up).

pull_comps

Real sale and rental comparables for a subject address, with value estimates and ZIP-level market stats.

str_market_data

Nearby short-term-rental listings with ADR, occupancy, and revenue quartiles for any coordinates.

Full tool documentation β†’ Β· keys ride the connection as an X-API-Key header (setup)

Listed on the Official MCP Registry (io.reizer/deal-analysis) and Smithery.

Why not just ask the AI directly?

REIzer MCPGeneric listing-scraper MCPsYour AI alone
Purpose-built underwriting tools6, covering the full deal lifecycle2–4, returning raw page datanone
Strategies coveredLTR Β· STR Β· BRRRR Β· flip Β· multifamily Β· new constructionβ€”β€”
Mathdeterministic engine β€” same inputs, same answernone (data only)LLM arithmetic, varies run to run
Stress testing7 scenarios per rental deal, 12 for flipsβ€”β€”
Max-offer solverbinary search on the real engineβ€”rule-of-thumb guesses
Comps datalicensed sale + rent + STR feedsscraped, fragile, often against ToSnone / hallucinated
Shareable output30-day report page per analysisβ€”chat text only
Setupone commandvaries, often needs proxiesβ€”
Costanalysis free Β· data metered by planoften paid proxy creditsfree, but unverified

Dive deeper

FAQ

Is it really free?

The four analysis tools, yes β€” no key, no account. Anonymous callers share a ceiling of 5 requests/min and 60/day per IP (cloud-hosted AI clients call from a vendor IP, so that pool is shared). An API key on a paid plan meters you per account instead β€” 1,000 analyses/day on Investor, 5,000 on Pro, 20,000 on Elite β€” and unlocks pull_comps and str_market_data against your plan's monthly data allowances.

Does REIzer's server run its own AI?

No. The server is a deterministic underwriting engine β€” the same math as the REIzer analyzers, documented under Methodology. Your AI does the talking; REIzer does the arithmetic. Same inputs, same answer, every time.

How is this different from asking my AI to "analyze a deal"?

Without tools, an LLM estimates the math and the market β€” plausibly, but unverifiably. With REIzer connected, it calls a real engine and real licensed data, and hands you a report page you can check line by line.

Prefer raw REST?

Everything the MCP server does is also a plain JSON API β€” see the API reference. Same engine, same free tier.

Give your AI real underwriting.

One command. Six tools. Six strategies. Deterministic math your AI can't hallucinate β€” free to start.

$0core analysis, no account
96/100the demo deal above β€” real output
1command to connect