---
name: boon-property-research
version: "0.1.0"
description: "Use for a guided US property-record workflow; ask for the address and decision first, then check Locus coverage before free, cited research."
license: MIT
triggers:
  - "research this property"
  - "check this address before I decide"
  - "look up zoning, permits, taxes, or development records"
  - "run property due diligence"
  - "what should I check about this address"
tags:
  - property-research
  - public-records
  - locus
  - evidence-led
metadata:
  workflow: guided
  provider: locus
  payment: free-first
  requiresHumanApprovalForPaidCalls: true
---

Use this as a **guided workflow**, not as a completed property report or a
claimed test. It produces a bounded research plan and, only after live calls,
an evidence-labeled summary. It never gives a valuation, eligibility,
approval, legal outcome, title result, utility-service determination, or safety
label.

## Required first questions

Ask both questions before looking anything up or making a tool call:

1. **What exact property address should I research?** Keep the user's exact
   address string. If it could resolve to more than one jurisdiction, ask the
   user to choose. Do not append a city, county, ZIP, or state that the user
   did not provide.
2. **What decision will this research support?** Ask for the decision, such as
   comparing a purchase, planning a renovation, checking a permit path, or
   reviewing a tax question. This scopes the records without turning the
   answer into a recommendation.

If either answer is missing, ask and stop. Never assume an address, invent a
decision, scrape a portal, or call an endpoint to fill the gap.

## Locus-first sequence

1. Load the current Locus skill from
   `https://docs.locus.report/skill.md`. Do not copy an old tool definition
   into this workflow.
2. Load the current free tool catalog from
   `https://api.locus.report/tools/list`. The catalog is authoritative for
   names, schemas, coverage, and free or paid rails. The remote Locus MCP
   endpoint is `https://mcp.locus.report/mcp` when the client supports MCP.
3. Call `locus_execute` with the exact user address:

   ```json
   {
     "name": "locus_lane_availability",
     "arguments": {
       "place": "SUPPLIED_EXACT_ADDRESS",
       "detailLevel": "full"
     }
   }
   ```

   Replace `SUPPLIED_EXACT_ADDRESS` with the exact address the user supplied.
   Never send the placeholder or invent a different address.

   Lane availability is a capability and coverage map. It is **not a finding**,
   a parcel result, or proof that a property has or lacks a condition.
4. From the live availability result, choose only relevant **free** tools. For
   a property decision this may include parcel facts, zoning, permits, tax
   records, and development or planning records. Respect each tool's input
   schema, jurisdiction, coverage, and status. Prefer the narrowest tool that
   answers the stated decision.
5. Read each returned source, record date, jurisdiction, status, and caveat.
   Preserve `candidate`, `reported`, `partial`, `stale`, `unavailable`, and
   `not_verified` states. Missing records or an unavailable lane are not proof
   of no record, no issue, approval, eligibility, or service.
6. Stop after the relevant free calls. Do not buy a paid report, start a
   subscription, or set monitoring. If the user later asks for an optional
   paid call, show the exact live challenge price, recipient wallet, network,
   asset, and scope first. Ask for explicit approval of that exact call. Never
   use a cached recipient, guessed wallet, or fixed payment promise.

## Output contract

Return a short, source-linked research handoff with:

- the exact address string and the decision the user supplied;
- the resolved place or parcel, with the resolution method and jurisdiction;
- records grouped by question, with source URL and observed date;
- an evidence label for every material statement;
- coverage limits, stale or missing sources, and the next user-confirmed step;
- a clear statement that this is guided research, not a completed property
  report or a claimed test.

Use plain language. Say what a record reports. Do not score, rank, predict,
value, screen, or label a person, property, block, or neighborhood as
safe/unsafe. Do not infer utility service, tax eligibility, permit approval,
development capacity, ownership, title, or legal outcome from a candidate
record or from missing data. Keep private personal information out of notes
and output unless it is necessary for the user's stated decision.
