The monthly supply scan every local investor should run before writing an offer. A Realtor skill teaches Claude which of the 14 Monthly Housing Inventory columns answer list-price, velocity, and price-cut questions — and how to read them against state + national benchmarks, not in isolation.

Reference monthly housing-inventory data from Realtor (Realtor.com Research, Monthly Housing Inventory feed) for real estate investment decisions. Use this skill whenever a user asks about Realtor.com data, median listing price, active listings, days on market, pending listings, pending ratio, new listings, total listings, price increases, price reductions, price-cut share, median price per square foot, average listing price, median square feet, listing supply, county-grain inventory, native YoY change, county-vs-state-vs-national benchmarks, monthly housing inventory CSV, RDC inventory metrics, or which Realtor columns inform a residential RE decision. Maps 14 Realtor Monthly Housing Inventory columns across 5 groupings (prices, supply, demand_velocity, pricing_action, mix) to investor use-cases. Progressive disclosure — decision playbooks and the full 14-dataset reference live in `references/`.

Realtor for Real Estate

What is Realtor

Realtor is Realtor.com's research arm — realtor.com/research/data. Free, no signup, CSV downloads only (no API), refreshed monthly with the Monthly Housing Inventory feed plus a few weekly and hotness reports we don't track here. For US residential real estate, it's the authoritative single stop for county-grain listing-side reads — list prices, active and pending counts, days on market, and price-cut behavior — pulled from the MLS pipeline that powers the consumer site.

Realtor's Monthly Housing Inventory ships at national, state, metro, county, and ZIP grain — and county is the finest cut that fills reliably across all 14 columns. For tract-level demand reads (income, tenure, vacancy), pair with Census ACS. For weekly velocity or sale-side pricing, pair with Redfin. For long-history ZIP-level home values, pair with Zillow's ZHVI.

Bundled resources

Progressive disclosure rule: start with this file and the category table below. Read references/metrics.md only when a specific column or csv_column is actually needed. Read the YAML only when the markdown is insufficient.

Getting an API key

1. No API key required. The Monthly Housing Inventory feed ships as public CSVs on S3 — pull them with pandas.read_csv or requests.
2. No environment variable to set. The catalog's csv_url fields are the URLs to hit directly.
3. No dedicated client. pip install pandas covers every pull; requests works too if you want raw bytes.
4. No rate limit published. The CSVs are static monthly snapshots — pull once a month, cache locally, don't hammer the host.

Minimal Python example

import pandas as pd
 
# Median listing price — Hillsborough County, FL
url = "https://econdata.s3-us-west-2.amazonaws.com/Reports/Core/RDC_Inventory_Core_Metrics_County_History.csv"
df = pd.read_csv(url)
df = df[df["county_fips"] == 12057]
df["date"] = pd.to_datetime(df["month_date_yyyymm"], format="%Y%m") + pd.offsets.MonthEnd(0)
df = df.sort_values("date")
print(df[["date", "median_listing_price", "median_listing_price_yy"]].tail(12))

Swap median_listing_price for any csv_column from references/metrics.md. Swap _County_ for _State_ (filter on state_id) or _Country_ (no filter — one row per month) for the state/national CSVs.

The 5 groupings

Total: 14 columns. See references/metrics.md for the full list with csv_columns and rationale.

Decision playbooks

Concrete "the user asked X — here's the column set to pull." Use these as starting points; extend when the question demands it. If a playbook names a column, references/metrics.md has the verbatim rationale.

"Is supply tightening or loosening here?" → active_listing_count, new_listing_count, total_listing_count, median_days_on_market Active is the live snapshot; new is the inflow; total = active + pending (separates real supply from churn); DOM is the absorption read. Run them together against state + national to see if local supply is moving with the cycle or against it.

"Where's negotiating leverage right now?" → price_reduced_share, price_increased_share, pending_ratio, median_days_on_market Share-form price-cut and price-increase columns smooth out inventory swings; pending ratio above 1.0 says deals-in-motion outnumber for-sale homes (hot market). DOM stretching while cuts spread = buyer leverage building.

"Are Tampa prices a premium or a discount?" → median_listing_price, median_listing_price_per_square_foot, average_listing_price, median_square_feet Median is the headline; per-sqft controls for size-mix shifts; average vs. median gap reads the luxury-tail influence; median square feet flags whether a mix change is driving the headline. Benchmark all four against FL + US.

"Is demand running ahead of or behind closings?" → pending_listing_count, pending_ratio, median_days_on_market, new_listing_count Pending count is the 30–60 day leading indicator of closed sales. Pending ratio normalizes for inventory. DOM and new listings frame whether the pending build is real demand or just fresh supply moving fast.

"Are sellers giving up pricing power?" → price_reduced_count, price_reduced_share, price_increased_count, price_increased_share Watch the share lines for cycle turns — the count versions move with raw inventory. When Hillsborough's reduced-share runs above FL and US while increased-share runs below, sellers are blinking first locally.

"Quick monthly Tampa pulse — three columns to scan?" → median_listing_price, active_listing_count, median_days_on_market The three TiRE surfaces as the headline Realtor metrics. Price level, supply level, velocity — every other column in the catalog adds nuance to one of these three.

Gotchas

• CSV-only, no API. The Monthly Housing Inventory feed is a public S3 CSV. No tokens, no JSON endpoint, no historical snapshots beyond the rolling file. Cache locally if you need a point-in-time read.
• Three separate CSVs by grain. County, state, and national each live in their own file (_County_, _State_, _Country_ in the URL stem). The fetcher in agents/sources/realtor.py does the substitution; if you're pulling raw, swap the stem yourself.
• Native YoY columns. Every metric ships with a paired _yy column (decimal float — multiply by 100 for percent). Use these directly per decision #72; don't recompute YoY off the level series.
• Country CSV has no filter column. It's one row per month nationally. The county and state CSVs need filtering on county_fips (int) or state_id (string like "FL").
• County FIPS as integer. county_fips in the CSV is an int, not a zero-padded string. Hillsborough is 12057 (not "12057"); cast accordingly when filtering.
• Monthly cadence, ~3 week lag. The previous month's data drops mid-month. Don't expect intra-month freshness — Realtor isn't Redfin's weekly feed.
• History starts 2016-07. All three CSVs ship 2016-07 onward. ~117 months as of 2026-04. Anything earlier needs a different source.
• Two dropped Sprint 4 datasets are gone for good. realtor_hotness_monthly_county and realtor_weekly_national were cut in Sprint 9 (decision #73) — don't propose them. Their columns are either redundant with Monthly Inventory or national-only YoY-only, neither of which earns a /metric page.

Source of truth

This skill is derived from TiRE's canonical catalog at agents/data/realtor-datasets.yaml. The YAML is hand-curated; every entry has a why_it_matters rationale that feeds both this skill and TiRE's research agents.

When the upstream catalog changes, regenerate references/metrics.md and references/realtor-datasets.yaml. v1 is hand-curated; build script deferred until drift is observed.