Skip to main content

​
Short Answer

Yes, but it is not a button. We do not offer a “click once and it is configured” integration in the traditional sense. Different models use different protocols, parameters and authentication, so such a button could only cover basic chat and would not solve real needs. What we offer instead works better: give the documentation to your AI coding assistant and let it do the integration. You only register an account and copy a key. Choosing the model, filling in the base URL, writing the code and debugging can all be left to the AI.

Let an AI integrate for you

The getting-started page has a copyable prompt for Codex, Claude Code and Cursor.

AI Developer Kit

A skill, a CLI, and a contract plus model registry for coding agents, each with its own prompt.

​
Four Ways to Use It

1

The site-wide skill (recommended)

Have your agent run npx skills add https://docs.apiyi.com to install the APIYI skill. If that fails, have it read https://docs.apiyi.com/skill.md directly.That file is written for machines: endpoints, authentication, model-naming rules, known pitfalls and a verification checklist. Reading it gives the agent all the background it needs.
2

Send one page to an AI

Every documentation page has a Copy page button at the top right. The arrow next to it also offers “Open in ChatGPT / Claude / Perplexity / Google AI Studio”.When you hit a problem with a specific model, open that page, click Copy page, and send it together with your error message. This is the fastest debugging path.
3

Connect as an MCP server

Add https://docs.apiyi.com/mcp as an MCP server and your agent can search the latest content of this site whenever it needs to, without you feeding it pages by hand.
4

Command line, zero install

If you do not want to write code, run npx apiyi@latest check in a terminal (Node 18+). It checks the key, probes the node and lists models; then npx apiyi@latest chat "Hello" -m gpt-5.4-mini sends the first message. The command table is in the AI Developer Kit.
Why this is better: a one-click button can only cover a fixed set of scenarios. An AI reads your project’s actual stack, writes code that runs, and handles timeouts and retries along the way, which a button never could.

​
Why Not a One-Click Button?

Model integration differs in several ways:
  • Different protocols: OpenAI, Claude, Gemini and others use different API protocols
  • Different parameters: request fields and response shapes vary across models
  • Different authentication: auth headers and their locations are not standardized
  • Different capabilities: Function Calling, Prompt Caching and Web Search are implemented differently on each model
A forced one-click flow would usually only cover basic chat, which is not enough for real use.
You still register the account yourself. There is no API for an agent to create an account or a key on its own. Accounts and billing involve identity and risk control, so a human has to do that step. Everything from “I have a key” to “the code works” can be handed to the AI.

Getting Started

Two paths: let an AI integrate, or do it yourself.

How to Choose a Model

Pick the most suitable AI model for your use case.

How to Configure the Base URL

How to connect APIYI in various clients.

How to View Call Logs

Check API call records and balance consumption.