v0.5.2 release - Contributors, Sponsors and Enquiries are most welcome 😌

Quick Start Guide

Get up and running with AgentSea in minutes. Build your first AI agent with just a few lines of code.

Installation

Install AgentSea using your preferred package manager:

bash
pnpm add @lov3kaizen/agentsea-core
npm install @lov3kaizen/agentsea-core
yarn add @lov3kaizen/agentsea-core

Basic Agent

Create your first agent in just a few lines of code:

typescript
import { Agent, AnthropicProvider, ToolRegistry, BufferMemory, calculatorTool } from '@lov3kaizen/agentsea-core';
const provider = new AnthropicProvider(process.env.ANTHROPIC_API_KEY);
const toolRegistry = new ToolRegistry();
toolRegistry.register(calculatorTool);
const agent = new Agent({
  name: 'my-assistant',
  model: 'claude-sonnet-4-20250514',
  provider: 'anthropic',
  systemPrompt: 'You are a helpful assistant.',
  tools: [calculatorTool],
}, provider, toolRegistry, new BufferMemory(50));
const response = await agent.execute('What is 42 * 58?', context);

Using Local Models (Privacy & Cost-Free)

Run agents completely locally with Ollama - perfect for privacy-sensitive applications, offline development, or eliminating API costs:

typescript
import { Agent, OllamaProvider, ToolRegistry, BufferMemory } from '@lov3kaizen/agentsea-core';
// No API key needed - runs 100% locally!
const provider = new OllamaProvider({
  baseUrl: 'http://localhost:11434',
  model: 'llama3.2' // or mistral, gemma2, qwen2.5, etc.
});
const agent = new Agent({
  name: 'local-assistant',
  model: 'llama3.2',
  provider: 'ollama',
  systemPrompt: 'You are a helpful assistant running locally.',
}, provider, new ToolRegistry(), new BufferMemory(50));
const response = await agent.execute('Hello!', context);

🦙 Get Started with Ollama

Install Ollama in seconds:

bash
curl -fsSL https://ollama.com/install.sh | sh
ollama pull llama3.2
→ Complete guide to local & open source providers

MCP Integration

Connect to MCP servers to extend your agent with external tools:

typescript
import { Agent, AnthropicProvider, ToolRegistry, MCPRegistry } from '@lov3kaizen/agentsea-core';
const mcpRegistry = new MCPRegistry();
await mcpRegistry.addServer({
  name: 'filesystem',
  command: 'npx',
  args: ['-y', '@modelcontextprotocol/server-filesystem', '/tmp'],
});
const mcpTools = mcpRegistry.getTools();
const toolRegistry = new ToolRegistry();
toolRegistry.registerMany(mcpTools);
const agent = new Agent(config, new AnthropicProvider(), toolRegistry);
const response = await agent.execute('List the files in /tmp', context);

Multi-Agent Workflows

Orchestrate multiple agents for complex tasks:

typescript
import { WorkflowFactory, AnthropicProvider, ToolRegistry } from '@lov3kaizen/agentsea-core';
const workflow = WorkflowFactory.create({
  name: 'research-workflow',
  type: 'sequential',
  agents: [
    { name: 'researcher', systemPrompt: 'Research information.' },
    { name: 'writer', systemPrompt: 'Write a summary.' },
  ],
}, new AnthropicProvider(), new ToolRegistry());
const result = await workflow.execute('Research AI agents', context);

Next Steps

💡 Tip

Check out the examples page for complete, runnable examples covering all features.