Produce an area report in ten minutes with the Insights skill

An area report used to be an afternoon: pull the index, find the volumes, count the listings, look up the agents, write it up. A skill does it in one pass, from the same tools, in the same order, with every figure cited. This guide runs the area-insights skill for one district and explains what comes back.

What a skill is

A skill is a short markdown file that tells an agent how to produce a piece of work: which MarketCode tools to call, in what order, what each costs, and how to lay out the result. It is not a prompt template. The agent reads it and does the work, and the report comes out the same shape every time. The skills are public, in the marketcode-plugin repository, and the skills page lists them with what each produces.

Step 1: install the plugin (Claude Code) or paste the skill (anything else)

In Claude Code, two commands add the MCP server and every skill:

/plugin marketplace add robosapien-ai/marketcode-plugin

/plugin install marketcode@marketcode

Then /mcp, choose MarketCode, Authenticate, and sign in with your MarketCode account.

In Claude, ChatGPT, Codex or Cursor, connect the MCP server first (steps for each client), then open the area-insights skill, copy its text and paste it at the top of your conversation. The agent treats it as instructions.

Step 2: name the area

Run the area-insights skill for Lewisham.

Or a postcode district: "for SE13". Or a comparison: "for Lewisham, Greenwich and Southwark". Local authority names and postcode districts both work; the skill resolves whichever you give with area_resolve, which also tells it the area's parents and neighbours.

Step 3: read the report

The skill calls a dozen free tools and lays the answers out in a fixed order:

• Headline facts from market_facts: the median sale price, price per square metre and turnover for the last settled month, by property type and bedroom count, each with an evidence grade so a cohort with three sales is not read as a market.
• The sales index from market_index_series: a repeat-sales index, so a change in what sold cannot move it, with its twelve-month and five-year change and a marked forecast tail.
• Asking prices and rents from asking_price_index_series and asking_rent_index_series: hedonic indices from live listings, which lead the sold index by months.
• Volumes from market_volume_series: registered sales per month, with the months still filling at HM Land Registry marked provisional rather than read as a collapse.
• Live stock from listing_stock_series: how many homes are for sale today, the asking median, how old the stock is, and the share reduced.
• Agents from agent_activity: who is listing and selling in the area.
• Auctions from auction_stats: lots offered, sold and the discount to guide, by quarter.
• Neighbours from market_ranking: the same headline figures for the surrounding areas, ranked.

Every number carries the tool it came from and the period it refers to. Where a series is quarterly rather than monthly, or the index is the national series standing in for a thin local one, the report says so in the line.

Step 4: what it cost

Nothing. Every tool in this skill is free, and the skill states that at the top so the agent never has to guess. If you run it over the REST API instead, it costs the requests and no credits.

The same figures, without an assistant

Each local authority and postcode district also has a page under Markets with the same series drawn from the same tools, updated daily. The skill is for when you want the comparison written up, or a question the page does not ask.