Agent Skills
Skills are packaged instructions an AI agent loads on demand — they teach the agent how a tool works so you don't have to explain it in every prompt. The RealtyAPI skill teaches Claude Code (and any skills-compatible agent) our base URLs, authentication, and how to discover the right endpoint for a task.
Install
One command, from your project directory:
npx skills add realtyapi/realtyapi-skills
The source is open on GitHub: realtyapi/realtyapi-skills
What your agent learns
Gateway shape & auth
One subdomain per provider (https://{provider}.realtyapi.io), API key in the x-realtyapi-key header, JSON responses.
Endpoint discovery
How to find the right provider and endpoint for a task using llms.txt and each provider’s OpenAPI spec — no guessing at paths.
Integration patterns
How to wire RealtyAPI into your codebase: pagination, parameter handling, and error-code conventions.
Use it
After installing, just describe what you want — the agent activates the skill automatically when a task involves real estate data:
Add a function to my app that fetches Redfin listings for a given city.
Build a script that pulls Rightmove sold prices for a postcode and charts the trend.
Wire up a nightly job that checks new Apartments.com rentals in Denver.
The agent will ask for your API key when it needs one (or read it from your environment). Get a free key from the dashboard — 250 requests/month, no card required.
Skill or MCP server?
Use the skill when your agent is writing code that integrates RealtyAPI into your project. Use the MCP server when you want the agent to fetch live data itself during a conversation. They work well together — the skill for building, MCP for querying — and both use the same x-realtyapi-key.