Score UK areas, query signals in natural language, watch portfolios for material change, and generate audience-shaped briefs, all inline in your AI workflow. Eleven tools across six product surfaces. The server is distributed as an npm package and spawned by your MCP client over stdio.
One API key, one config edit. The client will spawn the server as a subprocess and the tools appear in your next conversation.
Get an API key from your dashboard. Keys start with oga_.
Add the server to your MCP client config.
Edit ~/Library/Application Support/Claude/claude_desktop_config.json on macOS, or %APPDATA%\Claude\claude_desktop_config.json on Windows. Add the onegoodarea server:
{
"mcpServers": {
"onegoodarea": {
"command": "npx",
"args": ["-y", "@oga-mcp/server"],
"env": {
"OOGA_API_KEY": "oga_..."
}
}
}
}Restart Claude Desktop. The eleven tools appear when you start a conversation about UK locations.
Add to .cursor/mcp.json in your project (or the global Cursor MCP config):
{
"mcpServers": {
"onegoodarea": {
"command": "npx",
"args": ["-y", "@oga-mcp/server"],
"env": { "OOGA_API_KEY": "oga_..." }
}
}
}Every tool calls the same v1 API under the hood. The MCP server is a thin protocol bridge. Auth, rate-limits, quota, methodology pinning, all enforced on the API side. Every brief and summary the LLM sees is composed server-side from real engine state. No client-side text synthesis.
Composite scoring per area for one of four decision presets.
score_postcode(area, preset)Score a UK postcode or place name for a preset (moving, business, investing, research). Returns a 0-100 score, five weighted dimensions with engine-grounded reasoning and confidence, a server-composed summary, recommendations, and data sources.
compare_postcodes(areas[], preset)Score 2-8 areas side-by-side for the same preset. Returns a sorted comparison table with per-area summaries. Partial failures surface inline rather than failing the call.
Raw addressable signals: the primitive underneath every other product.
get_area_signals(area)Full signals catalog for an area across all seven categories (crime, deprivation, property, schools, amenities, transport, environment). Each signal carries value + unit, percentile when store-backed, confidence with engine-grounded reason, source attribution, and observation period.
get_signals_by_category(area, category)Same signal shape as get_area_signals, narrowed to one category. Use when the LLM needs to focus on a single data domain.
Natural-language query plane + peer discovery over the moat.
find_areas(question)Ask in natural language. A planner translates the question into a typed plan (one of seven ops: rank_areas, get_area, score_area, compare_areas, find_peers, find_insights, find_forecast); the database executes it. The response carries the emitted plan + results so every answer is reproducible.
find_peers(area, k?)k-nearest-neighbour peers for a UK area by normalized signal values. Returns the target's geo_code + signals_used + a ranked peers list with distance (0 = identical, 1 = maximally distant) and n_dims_used.
Portfolio tracking and material-change detection.
watch_portfolio(name, areas[])Set up a Monitor portfolio in one step: creates the portfolio and adds the tracked areas. Returns the new portfolio_id and the area list. If the add step fails after the create, the response surfaces the partial state so the LLM can act.
get_portfolio_changes(portfolio_id, threshold_pct?, baseline?, min_transactions?)Check a portfolio for material signal changes between two time-series periods. Returns scope, counts, and a per-area table of material moves with direction, from/to values, delta, and percent change. Probe calls don't fire customer webhooks.
One audience-shaped advisory document per area. The wow-factor composite.
area_brief(area, audience)Audience ∈ {lender, insurer, retailer, investor}. Fans out to the full signals catalog + the audience's scoring preset (with explain mode), then renders an audience-specific markdown brief: overall verdict, audience-relevant dimensions, audience-relevant signals with provenance, recommendations, data sources. Every value is real engine output.
Static lookups, no quota cost.
methodology_for(dimension)Methodology for any scoring dimension: data source, scoring function summary, per-preset weights. Useful for procurement and model-risk review.
engine_version((no args))Current engine version, release date, and changelog excerpt. The live engine version is also stamped on every score_postcode and get_area_signals response.
For local testing against a development API instance, point the server at any base URL via OOGA_API_BASE. The server validates the key prefix but does not validate the host.
{
"mcpServers": {
"onegoodarea-dev": {
"command": "npx",
"args": ["-y", "@oga-mcp/server"],
"env": {
"OOGA_API_KEY": "oga_dev",
"OOGA_API_BASE": "http://localhost:4000"
}
}
}
}Open source, no per-call MCP fee. Tool invocations make real calls to the OneGoodArea API and consume your account quota the same way any other integration does. Current tiers and any MCP-specific terms live on the pricing page.
Get a key, paste a config block, restart your client. The eleven tools appear in your next conversation.