AI Decision System
Replayable, offline-verifiable records for AI-assisted go/no-go decisions: experiment promotions, typed model answers and human-approved actions.
Official MCP server for the Location Drive API — connect Claude, Cursor, or any MCP client to 307.8M places in 250 countries and territories, 254.2M graded building polygons, and brand intelligence.
Developer toolsTypeScript · HTMLGitHub @LocationdriveOnlineOwner unconfirmed
Have you tried it? One tap is enough. No sign-in needed.
No updates detected yet. GitHub releases will appear here.
Owners of public repositories under a personal GitHub account can sign in to verify ownership and manage the introduction themselves. No administrator approval is needed.
Sign in, generate a verification file, and commit it to this repository’s default branch within 30 minutes. Files on other branches or forks do not count. Then check the file below to manage the project without administrator approval. Keep the file to retain your access.
If needed, you will go to sign-in. After signing in, return here and press your chosen verification button again.
Run /nomorevibe verify in your project folder to verify ownership and manage the introduction yourself.
Verify in one minuteAsk about other verification options
Loading the full README. The saved excerpt is shown below.
Query the Location Drive location-data platform from Claude, Cursor, or any
MCP-compatible AI client: global POI search, graded building-footprint
polygons, brand intelligence, and premium data packs (EV charging, fuel
prices, full menus, multi-site hotel rates, multi-year review history) — using
your own Location Drive API key, so your plan, quotas, and usage dashboard
apply exactly as with direct API calls.
The dataset: 307.8M places across 250 countries and territories, 254.2M A–E
graded building polygons, 97 structured fields on every plan — Free included.Read the full README ↗Query the Location Drive location-data platform from Claude, Cursor, or any MCP-compatible AI client: global POI search, graded building-footprint polygons, brand intelligence, and premium data packs (EV charging, fuel prices, full menus, multi-site hotel rates, multi-year review history) — using your own Location Drive API key, so your plan, quotas, and usage dashboard apply exactly as with direct API calls.
The dataset: 307.8M places across 250 countries and territories, 254.2M A–E graded building polygons, 97 structured fields on every plan — Free included.
| Tool | What it does | Cost |
|---|---|---|
search_places | Search the global POI database by name/keyword — ranked by prominence for general terms, by similarity for specific names | limit per page (default 20); max_results auto-pages and sums the cost |
find_nearby | Distance-sorted places around a coordinate (adds distance_m) | limit per page (default 20); max_results auto-pages |
brand_locations | Every location of a brand — one country or global — with the total | limit per page (default 20); max_results auto-pages |
get_place_details | Full record by ID; request any of the 97 fields via fields — premium packs (menus, hotel rates, EV connectors, fuel, review text) included on every plan; data_note when a requested field has no data for that place | 1 |
get_place_context | LLM-ready summary + structured data (sentiment keywords, not review text) for one place | 1 |
get_review_history | Merged multi-year review text history, duplicates removed, plus sentiment summary | 1 |
scan_reviews | Keyword scan of review text across up to 40 places near a point — matching snippets per place; neighborhood scale | the nearby search (its limit) + 1 per scanned place, reported as api_calls_counted |
get_hotel_rates | Compare room rates across booking sites for one hotel | 1 |
get_building_polygon | GeoJSON footprint with A–E accuracy grade | 1 |
find_building_at_point | Reverse lookup: which building contains this coordinate | 1 |
brand_footprint | Store counts, open-now, coverage, avg rating for a brand | 1 (summary) |
data_coverage | Per-country POI counts and polygon coverage stats | 1 |
All plans receive all 97 fields. Usage is counted per place requested: list
tools cost their limit (default 20); single-place and summary tools cost 1.
Free keys: 5,000 API calls/month, 20 results per call. A list request is
charged its limit whatever it returns, so the list tools never send more
than 20 unless you set limit; max_results (≤ 1,000) pages automatically
and returns the summed api_calls_counted.
Get an API key at locationdrive.com → Dashboard → API Keys.
Claude Desktop — add to claude_desktop_config.json:
{
"mcpServers": {
"locationdrive": {
"command": "npx",
"args": ["-y", "@locationdrive/mcp"],
"env": { "LOCATIONDRIVE_API_KEY": "ld_live_YOUR_KEY" }
}
}
}
Claude Code:
claude mcp add locationdrive -e LOCATIONDRIVE_API_KEY=ld_live_YOUR_KEY -- npx -y @locationdrive/mcp
Cursor / other MCP clients: command npx -y @locationdrive/mcp with the
LOCATIONDRIVE_API_KEY environment variable.
The same server runs hosted at https://mcp.locationdrive.com/mcp
(Streamable HTTP) — no install needed. Authenticate with your key as a
Bearer header.
Claude Code:
claude mcp add --transport http locationdrive https://mcp.locationdrive.com/mcp \
--header "Authorization: Bearer ld_live_YOUR_KEY"
Cursor — add to .cursor/mcp.json:
{
"mcpServers": {
"locationdrive": {
"url": "https://mcp.locationdrive.com/mcp",
"headers": { "Authorization": "Bearer ld_live_YOUR_KEY" }
}
}
}
Other clients: URL https://mcp.locationdrive.com/mcp with header
Authorization: Bearer ld_live_YOUR_KEY.
Ask your AI client things like:
reviews (current crawl),
historical_reviews (earlier crawls), and get_review_history (both merged,
de-duplicated). When a requested field comes back missing, get_place_details
returns a data_note: nothing was stripped, that place has no data for it.
get_place_context carries review_sentiments keyword counts, not review text.scan_reviews — it runs the nearby search, reads up to 40 places' review
histories (one API request each; ld_test_ keys are capped at 10 places),
and returns only the matching snippets.
Keyword matching is a coarse filter over user-submitted text: read the
snippets before drawing conclusions.limit per page (default 20, capped at the plan's results per request)
whatever the page returns, single-place and summary tools cost 1, and
scan_reviews costs the nearby search plus 1 per scanned place. Every tool
result includes api_calls_counted (from the response body, or the
X-API-Calls-Counted header when the body lacks it) — see
https://locationdrive.com/docs#api-call-counting.429 QUOTA_EXCEEDED means the key's monthly allowance is exhausted (the
refused request is not counted); it resets on the 1st at 00:00 UTC.ld_test_ keys work and return synthetic data without counting against
quotas — useful for trying the integration.LOCATIONDRIVE_TIMEOUT_MS overrides).npm install
npm run build # compiles src/ and typechecks api/
npm run inspect # MCP Inspector against the local build
LOCATIONDRIVE_API_KEY=ld_live_x node scripts/smoke.mjs # live API smoke test
MIT © Location Drive, LLC