agentnode-ts

A lightweight AI agent framework for TypeScript.

Features

• 🤖 Provider-independent Model interface
• 🔄 Automatic execution loop
• 🛠️ Tool registration and execution
• 🔁 Multiple tool calls per iteration
• 🧩 Designed for multiple language model providers
• 📦 Fully typed TypeScript API

Installation

npm install agentnode-ts

Quick Start

import {
  AgentNode,
  OpenAIModel,
} from "agentnode-ts";

const model = new OpenAIModel({
  model: "gpt-4.1-mini",
});

const agent = new AgentNode({
  model,
  instructions: "You are a helpful assistant.",
});

const response = await agent.run(
  "Explain what an AI agent is."
);

console.log(response.text);

Defining a Tool

import type { Tool } from "agentnode-ts";

export const getCurrentTimeTool: Tool = {
  name: "get_current_time",

  description: "Returns the current time for a given time zone.",

  inputSchema: {
    type: "object",
    properties: {
      timeZone: {
        type: "string",
      },
    },
    required: ["timeZone"],
  },

  async execute(input) {
    return {
      time: new Date().toISOString(),
    };
  },
};

Register the tool when creating the agent.

const agent = new AgentNode({
  model,
  tools: [getCurrentTimeTool],
});

The framework automatically:

• sends tool definitions to the language model
• executes requested tools
• updates the conversation context
• continues the execution loop until a final response is produced

Examples

Run the basic example:

npx tsx examples/basic.ts

Run the tool-calling example:

npx tsx examples/current-time/index.ts

Supported Providers

Current:

• OpenAI

Roadmap

This framework is under development. Some of the planned features include:

• Streaming responses
• Memory
• MCP support
• Structured output
• Multi-step planning
• Additional model providers

License

MIT