You are producing an area-level market brief: what the market is doing, how it compares with its neighbours, and what to watch. The deliverable is markdown with tables; every figure names the tool it came from.

Load marketcode-property-research first for the tool vocabulary and the wording rules.

Credit budget

Every tool below costs 0 credits. Say so, then proceed.

Inputs

An area: a postcode district (BS5), a town, a borough or a local authority. Optionally a bedroom band or property type focus, and a comparison set.

Phase A: evidence (parallel)

Resolve. area_resolve(area_name) once. Use the area_id it returns for the area_id tools and the area_code and area_type it returns for the series tools. If it returns both an area_id and an analytics_area_id, pass analytics_area_id where the tool asks for analytics.
Prices and rents. market_facts for the area with max_periods of 36 months: price, £/m², volume, rent, yield, stock, turnover, by bedroom band where available.
Indices. market_index_series (sales, with the forecast tail flagged), asking_price_index_series, asking_rent_index_series for the same area; commercial_index_series and commercial_rent_index_series only if the user asked about commercial.
Volumes. market_volume_series for the district over 36 months; market_volume_ranking to place it among its county or region.
Supply side. listing_market_series (new listings, reduction share, time to reduce, yields, agent concentration) and listing_stock_series (live stock by day). agent_activity(area_id) for who is transacting.
Stock. area_stock_profile(area_id), energy_profile(area_id).
Pipeline. planning_applications(location).
Auctions. auction_stats(postcode_district), auction_discount, and distressed_assets if the user is investing.
Peers. market_ranking for the parent area with the same metrics, so the subject sits in a league table of its neighbours. Never loop market_facts over areas to build a table; market_ranking is the tool for that.

Phase B: reading the evidence

Verdict first. One sentence that states the dominant story: "Buyer's market for flats, tight for houses, rents still rising."
Direction, not just level. For every headline metric give the latest value, the 12-month change and the 36-month change, and the peer median.
Thin data. market_facts carries evidence counts and provenance; if a cell rests on fewer than 20 transactions, say so and widen to the parent area for that metric.
Recent months are incomplete. Land Registry registrations lag by weeks to months. Treat the last two months of any sales series as partial and say so; the asking indices and listing stock are current.
Forecast tails are flagged in the index response; label them "projection" and never present them as observed.
Supply meets demand. Rising new listings with rising reduction share and lengthening time to reduce is softening; falling stock with falling reductions is tightening. Say which you see.
Auctions are the stress signal. Clearance rate, discount to the MarketCode estimate and repeat failures tell you where forced sellers are. About a quarter of lots carry an achieved price; read meta.coverage and say so.

Brief (markdown, in this order)

Verdict: one sentence, then three stance tags (sales, rentals, yield).
Headline metrics table: median price, £/m², gross yield, median rent, monthly volume, live stock, with 12-month and 36-month change and the peer median.
Price trends: sales index and asking-price index, 36 months, in a table or a chart spec the client can render; note divergence between asking and achieved.
Rental market: rent by bedroom band, asking-rent index, yield band.
Supply: new listings, reductions, time to reduce, live stock, most active agents.
Pipeline: recent applications by type and status, clustered.
Stock and energy: built form, tenure mix, EPC distribution.
Auctions: clearance, discount to estimate, repeat-failure lots.
Peers: the ranking table.
Watch-outs: three to five risks, each one line with the evidence.
Sources: one line per tool, with the period and the coverage note.

Rules

Numbers are raw and dated; percentages are percentages, not decimals.
Omit a section whose tool returned nothing, and say it was omitted; never fill a gap with a plausible number.
Plain English; explain what each finding means for a buyer, a seller or a landlord, not only what the data says.

Phase A: evidence (parallel)

Resolve. area_resolve(area_name) once. Use the area_id it returns for the area_id tools and the area_code and area_type it returns for the series tools. If it returns both an area_id and an analytics_area_id, pass analytics_area_id where the tool asks for analytics.

Prices and rents. market_facts for the area with max_periods of 36 months: price, £/m², volume, rent, yield, stock, turnover, by bedroom band where available.

Indices. market_index_series (sales, with the forecast tail flagged), asking_price_index_series, asking_rent_index_series for the same area; commercial_index_series and commercial_rent_index_series only if the user asked about commercial.

Volumes. market_volume_series for the district over 36 months; market_volume_ranking to place it among its county or region.

Supply side. listing_market_series (new listings, reduction share, time to reduce, yields, agent concentration) and listing_stock_series (live stock by day). agent_activity(area_id) for who is transacting.

Stock. area_stock_profile(area_id), energy_profile(area_id).

Pipeline. planning_applications(location).

Auctions. auction_stats(postcode_district), auction_discount, and distressed_assets if the user is investing.

Peers. market_ranking for the parent area with the same metrics, so the subject sits in a league table of its neighbours. Never loop market_facts over areas to build a table; market_ranking is the tool for that.

Phase B: reading the evidence

Verdict first. One sentence that states the dominant story: "Buyer's market for flats, tight for houses, rents still rising."

Direction, not just level. For every headline metric give the latest value, the 12-month change and the 36-month change, and the peer median.

Thin data. market_facts carries evidence counts and provenance; if a cell rests on fewer than 20 transactions, say so and widen to the parent area for that metric.

Recent months are incomplete. Land Registry registrations lag by weeks to months. Treat the last two months of any sales series as partial and say so; the asking indices and listing stock are current.

Forecast tails are flagged in the index response; label them "projection" and never present them as observed.

Supply meets demand. Rising new listings with rising reduction share and lengthening time to reduce is softening; falling stock with falling reductions is tightening. Say which you see.

Auctions are the stress signal. Clearance rate, discount to the MarketCode estimate and repeat failures tell you where forced sellers are. About a quarter of lots carry an achieved price; read meta.coverage and say so.

Brief (markdown, in this order)

Verdict: one sentence, then three stance tags (sales, rentals, yield).

Headline metrics table: median price, £/m², gross yield, median rent, monthly volume, live stock, with 12-month and 36-month change and the peer median.

Price trends: sales index and asking-price index, 36 months, in a table or a chart spec the client can render; note divergence between asking and achieved.

Rental market: rent by bedroom band, asking-rent index, yield band.

Supply: new listings, reductions, time to reduce, live stock, most active agents.

Pipeline: recent applications by type and status, clustered.

Stock and energy: built form, tenure mix, EPC distribution.

Auctions: clearance, discount to estimate, repeat-failure lots.

Peers: the ranking table.

Watch-outs: three to five risks, each one line with the evidence.

Sources: one line per tool, with the period and the coverage note.

Area insights

Credit budget

Inputs

Phase A: evidence (parallel)

Phase B: reading the evidence

Brief (markdown, in this order)

Rules

Area insights

Credit budget

Inputs

Phase A: evidence (parallel)

Phase B: reading the evidence

Brief (markdown, in this order)

Rules