@regolo/sdk

TypeScript/JavaScript SDK for the Regolo AI API.

Installation

npm install @regolo/sdk
# or
pnpm add @regolo/sdk

Quickstart

import { RegoloClient } from "@regolo/sdk";

const client = new RegoloClient({
  apiKey: "your-api-key",
  chatModel: "your-model",
});

// Single-turn chat
const [role, content] = await client.runChat("Hello, who are you?") as [string, string];
console.log(content);

// Streaming chat
const stream = await client.runChat("Tell me a story", { stream: true });
for await (const [, chunk] of stream as AsyncGenerator<[string, string]>) {
  process.stdout.write(chunk);
}

Configuration

You can configure global defaults via environment variables before calling tryLoadingFromEnv():

| Variable | Description | |---|---| | API_KEY | Default API key | | LLM | Default chat model | | IMAGE_GENERATION_MODEL | Default image generation model | | EMBEDDER_MODEL | Default embeddings model | | AUDIO_TRANSCRIPTION_MODEL | Default audio transcription model | | RERANKER_MODEL | Default reranker model |

import { tryLoadingFromEnv } from "@regolo/sdk";
tryLoadingFromEnv();

API Reference

new RegoloClient(options?)

Creates a new client instance.

| Option | Type | Description | |---|---|---| | apiKey | string \| null | API key | | chatModel | string \| null | Chat/completions model | | embedderModel | string \| null | Embeddings model | | imageGenerationModel | string \| null | Image generation model | | audioTranscriptionModel | string \| null | Audio transcription model | | rerankerModel | string \| null | Reranker model | | alternativeUrl | string \| null | Override the base API URL | | preExistentConversation | Conversation | Restore a previous conversation |

client.runChat(prompt?, options?)

Send a chat message and receive a response. Maintains conversation history automatically.

const result = await client.runChat("What is 2+2?");
// result: ["assistant", "4"]

// With options
const stream = await client.runChat("Explain gravity", {
  stream: true,
  maxTokens: 1024,
  temperature: 0.7,
});

client.completions(prompt, options?)

Text completions (non-chat).

const text = await client.completions("The capital of France is", { maxTokens: 10 });

RegoloClient.chatCompletions(messages, options?)

Static method for one-off chat completions without a client instance.

const result = await RegoloClient.chatCompletions(
  [{ role: "user", content: "Hello" }],
  { model: "your-model", apiKey: "your-key" },
);

client.embeddings(inputText, options?)

Generate text embeddings.

const embeddings = await client.embeddings("Hello world");
// or multiple inputs:
const embeddings = await client.embeddings(["text 1", "text 2"]);

client.rerank(query, documents, options?)

Rerank documents by relevance to a query.

const results = await client.rerank(
  "What is machine learning?",
  ["ML is...", "Cooking recipes...", "Deep learning is a subset of ML..."],
);

client.audioTranscription(file, options?)

Transcribe audio files.

const text = await client.audioTranscription("/path/to/audio.mp3");

client.createImage(prompt, options?)

Generate images from a text prompt.

const [imageBuffer] = await client.createImage("A sunset over the ocean") as Buffer[];

RegoloClient.getAvailableModels(apiKey, options?)

List available models.

const models = await RegoloClient.getAvailableModels("your-key");

License

MIT