Skip to main content

Using Skywalk API with AI Agents

· 3 min read

AI coding agents are changing how developers interact with APIs. Instead of writing every HTTP request by hand, you can describe what you need in plain English and let an AI agent handle the API calls, data parsing, and formatting for you.

Skywalk API is a natural fit for this pattern. Our structured JSON endpoints, simple authentication, and comprehensive property management data make it straightforward for AI agents to work with your AppFolio data.

What Are AI Coding Agents?​

AI coding agents are tools like Claude Code, OpenAI Codex CLI, Cursor, and Windsurf that can write and execute code on your behalf. With "tool use" capabilities, these agents can make API calls, process the results, and present them in whatever format you need.

Think of it as having a developer assistant that already knows how to use the Skywalk API. You say "show me my vacant units" and the agent handles the rest.

Why Skywalk API Works Well with AI Agents​

A few things make Skywalk API particularly well-suited for AI agent integration:

  • Simple REST + JSON: No complex GraphQL queries or SOAP envelopes. Every endpoint follows the same predictable pattern.
  • Single-header auth: Just pass X-API-Key in the request header. No OAuth flows or token refresh logic for the agent to manage.
  • Structured responses: Every response follows the same { meta, links, data } format, making it easy for an agent to parse and understand.
  • Comprehensive property data: 20+ endpoints covering properties, tenants, financials, leasing, maintenance, and more.

Quick Example​

Here's what it looks like to use Claude Code with Skywalk API. First, give the agent the API description:

Fetch https://api.skywalkapi.com/llms.txt for the Skywalk API reference.
My API key is in the SKYWALK_API_KEY environment variable.

Then ask it questions in plain English:

"List all my properties and how many units each has"

The agent will call GET /v1/properties, process the JSON response, and present a formatted summary. If the data needs refreshing, the agent automatically handles the polling pattern — waiting for meta.status to become "ok" before presenting results.

"Pull the cash flow statement for Sunnyside Villas for Q1 2026"

The agent calls GET /v1/cash-flow with the appropriate property and date filters, then formats the financial data in a readable way.

"Create a $45 utility charge for occupancy 12345"

The agent calls POST /v1/tenant-charge with the charge details and confirms the result.

Getting Started​

We've published a comprehensive guide with everything you need:

  • AI Agents Guide — Full setup instructions, tool definitions for Claude and OpenAI formats, code examples, and best practices
  • llms.txt — Machine-readable API description that AI agents can fetch directly
  • API Reference — Complete documentation of all endpoints and response formats

If you're already using an AI coding agent, pointing it at our llms.txt file is the fastest way to get started. The file contains everything the agent needs to understand the API and start making calls.

What's Next​

We're excited about the possibilities that AI agents open up for property management workflows. We'll continue improving our AI agent support based on feedback. If you have ideas or questions, reach out to us.