Quickstart¶

Get up and running with AiKit in a few minutes.

Installation¶

Install AiKit via npm, yarn, or pnpm.

npm install @codeworksh/aikit

Generate Models Catalog¶

AiKit uses a models.gen.json file to store metadata about available models, features, and context limits.

After installation, you need to generate this local models catalog so AiKit knows about all the available models. You can generate this manually using the CLI bundled with the package:

npx aikit modelgen

By default, this creates a models.gen.json file in your project root. AiKit will automatically read this file at runtime.

First Request¶

Here's how to resolve a model, build a conversation context, and get a response.

import { Type, type Static, type TSchema, llm, stream, Message } from "@codeworksh/aikit";

// Resolve a model from the built-in registry
const model = await llm("openai", "gpt-4o-mini");
if (!model) throw new Error("Model not found");

// Define tools with TypeBox schemas for type safety and validation
const getTimeTool = Message.defineTool({
    name: "get_time",
    description: "Get the current time",
    parameters: Type.Object({
        timezone: Type.Optional(Type.String({ description: "Optional timezone (e.g., Asia/Kolkata)" })),
    }),
});

// Setup conversation context
const context: Message.Context = {
    systemPrompt: "You are a helpful assistant.",
    messages: [
        Message.createUserMessage({
            role: "user",
            time: { created: Date.now() },
            parts: [{ type: "text", text: "What time is it in Bengaluru?" }],
        }),
    ],
    tools: [getTimeTool],
};

// Get complete response without streaming
const response = await stream.complete(model, context);

for (const part of response.parts) {
    if (part.type === "text") {
        console.log(part.text);
    } else if (part.type === "toolCall") {
        console.log(`Tool: ${part.name}(${JSON.stringify(part.arguments)})`);
    }
}

For handling real-time streams, check out the Streaming section.