Zerly AI provides the intelligence layer for the Zerly Developer Platform. It powers the Zerly VS Code extension, enabling instant codebase understanding through architecture maps and feature flow tracing.

To get started, authenticate with GitHub to receive your personal API key. All requests to the Zerly API must include this key in the Authorization header.

All API requests require a bearer token in the Authorization header. Your API key is generated when you sign in with GitHub.

Authorization: Bearer sk_zerly_xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx

Keys follow the format sk_zerly_<32-char-hex> and are stored as SHA-256 hashes — the raw key is never persisted. If you lose your key, use the dashboard to regenerate it.

The Zerly VS Code extension connects to your account using a deep link. Follow these steps to set it up:

1. Install the Zerly extension from the VS Code Marketplace.
2. In VS Code, run the command Zerly: Connect Account.
3. Sign in with GitHub in your browser.
4. Your API key is automatically sent to the extension via a vscode:// deep link.

Recommended production flow from extension: open https://zerly.tinobritty.me/connect?autoConnect=1&extensionId=publisher.extension&setupProviders=1 after sign-in to enable zero copy/paste auth and immediate handoff.

If your extension uses a different extension ID, pass extensionId so the browser opens the correct VS Code protocol target.

Alternatively, visit the /connect page to manually copy your key or trigger the deep link.

The Zerly API is OpenAI-compatible. Use it anywhere you would use the OpenAI chat completions endpoint with a custom base URL.

POST https://zerly.tinobritty.me/api/v1/chat/completions

curl -X POST https://zerly.tinobritty.me/api/v1/chat/completions \
  -H "Authorization: Bearer sk_zerly_xxx" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "zerly/zerlino-32b",
    "messages": [
      { "role": "user", "content": "Explain recursion" }
    ]
  }'

{
  "id": "chatcmpl-...",
  "object": "chat.completion",
  "choices": [{
    "message": { "role": "assistant", "content": "..." },
    "finish_reason": "stop"
  }]
}

The Zerly platform exposes a single, highly optimized model trained specifically for code architecture analysis and risk detection:

You must specify this exact model name in every request. Requests using other model identifiers will be rejected with a 400 Bad Request error.

To ensure fair access, all API keys are subject to the following rate limits:

Requests that exceed these limits receive a 429 Too Many Requests response. Limits reset on a rolling window basis.