# Planted Money

> Evidence-first screens for Cook and Lake County, Illinois. Planted Money reports listing,
> area, comp, and risk evidence for supported addresses. Property-specific values, rent-based
> economics, and investor target prices appear only when current evidence clears calibration
> gates; otherwise the response returns a range or refusal with a reason.
> Site: https://plantedmoney.ai · App: https://app.plantedmoney.ai

Planted Money is AI-native: it is exposed as a Model Context Protocol (MCP) server, so any
MCP-capable assistant can answer questions about a specific Cook or Lake County, Illinois
address by calling the tools below. Humans can browse the same data without an assistant at
[app.plantedmoney.ai](https://app.plantedmoney.ai).

## MCP endpoint

- URL: https://mcp.plantedmoney.ai/mcp
- Transport: streamable HTTP (JSON-RPC). Public beta, no key required, 30 calls/min per IP.
- Manifest: https://mcp.plantedmoney.ai/mcp/v2/manifest

## How to connect

Claude (claude.ai / desktop):
1. Settings -> Connectors -> Add custom connector.
2. Name it "Planted Money" and paste the MCP URL: https://mcp.plantedmoney.ai/mcp
3. Save, then start a chat and ask about any Cook or Lake County, Illinois address.

Claude Code (CLI):

```
claude mcp add --transport http planted https://mcp.plantedmoney.ai/mcp
```

## Tools (v2)

- `search_investment_inventory` — scored inventory by strategy percentile, filtered by bbox or named geography, `min_percentile`, `max_price`, `min_confidence`.
- `deep_underwrite_asset` — full pro-forma for an address, point, or property_id, plus a 5-year cashflow projection.
- `get_comps` — tiered ARV with full explainability: comps used, adjustments, dispersion, tier, confidence, and rejected comps with reasons.
- `get_tract_report` — every signal for a census tract, labeled.
- `market_pulse` — county-wide pulse: percentile-ranked leaders, Section-8 arbitrage ZIPs, reassessment-shock townships.
- `risk_report` — a separate 0-100 risk score and flags, never blended into opportunity.

Legacy aliases (deprecated): search_deals→search_investment_inventory, underwrite→deep_underwrite_asset, comps→get_comps. Aliases still work — each is marked `deprecated:true` with a `replacement` in the manifest, and removal ships with a changelog entry and at least 90 days' notice.

## REST contract

`GET /api/property?address=...` is the consumer contract: the same underwrite, reshaped into
plain language. Exact values, investor targets, rent economics, and a property verdict are
nullable and appear only when the evidence gate passes. Full schema:
https://api.plantedmoney.ai/api/openapi.json.

## Example prompts

- Use Planted Money to underwrite 7141 S Woodlawn Ave, Chicago
- Ask Planted Money whether the evidence supports an investor target price for 5817 S Kimbark Ave, Chicago at $325,000
- Have Planted Money list the things to know before buying 1638 N Campbell Ave, Chicago

## Refusals are honest

A missing answer is a real answer, not an error to route around. Unit count unknown returns
`units_required` — never a guessed duplex. A tract whose sales corpus isn't ingested yet (all
of Lake County, today) or a comp set that fails its confidence gates comes back suppressed —
`tract_not_ingested`, or `arv: null` with a `suppression_reason` — never an invented number.
Agents should relay the refusal and its reason, not retry-loop or substitute a guess.

## Machine-readable references

- OpenAPI: https://api.plantedmoney.ai/api/openapi.json
- MCP manifest: https://mcp.plantedmoney.ai/mcp/v2/manifest
- Valuation contract: https://api.plantedmoney.ai/api/spec/valuation-contract.json
- Scoring contract: https://api.plantedmoney.ai/api/spec/scoring-contract.json
- Constants registry: https://api.plantedmoney.ai/api/spec/constants-registry.json
- Error taxonomy: https://api.plantedmoney.ai/api/spec/error-taxonomy.json

## Disclosure

Planted Money is an automated valuation model — not an appraisal. Every number ships with a
method tier, a confidence score, and an as-of date; published error rates are on the live
accuracy scorecard: https://docs.plantedmoney.ai/methodology/accuracy/. Outputs may not be
used for tenant screening, credit, employment, or insurance decisions, or any purpose governed
by the Fair Credit Reporting Act (FCRA).