Getting started
Install SiteScout, add a Geoapify key, and run your first site-selection analysis.
On this page
SiteScout turns a plain-language question into a ranked shortlist of real sites, and every number in the report links back to the Geoapify API call that produced it.
Requirements
Node.js 20+ and a free Geoapify key from myprojects.geoapify.com. The free plan gives 3,000 credits a day and needs no card.
Install
git clone https://github.com/ducvuathuduc/sitescout-agent.git
cd sitescout-agentnpm ci
npm run buildnpm run build compiles the TypeScript sources to dist/, including the CLI and the MCP server.
Add your key
Copy .env.local.example to .env.local and put your key in it. The CLI loads that file with dotenv (see src/cli.ts), so you do not need to export anything.
GEOAPIFY_API_KEY=YOUR_KEYRun your first analysis
node dist/cli.js scout "Where should I open a specialty coffee shop in Thảo Điền?" --id my-coffeeThe run writes report.md, report.html, result.json and trace.jsonl to examples/my-coffee/. Each line of trace.jsonl is one API call: its endpoint without the key, status, latency, summary and credits.
Use it from Claude Code or Cursor
SiteScout ships a stdio MCP server with five tools. Register it in your editor and ask the same questions in natural language.
Next steps
MCP usage
How SiteScout uses the Geoapify MCP server, and how to use SiteScout's own five tools.
Demo proxy
How the public demo keeps the key secret and shares one daily budget.
Credits and limits
How each Geoapify call is priced, and the caps the demo enforces.
How it works
Plan, geocode, isolines, scoring — and where each step lives in the code.