LLM Agent Library

A minimalist, portable JavaScript/TypeScript library for building tooled LLM agents.

Features

• 🚀 Lightweight & Portable: Built in TypeScript, works in any JavaScript environment
• 🔧 Extensible Capabilities: Add custom tools and abilities to your agent
• 🎯 LLM Agnostic: Works with any LLM provider (OpenAI, Anthropic, etc.)
• 🎨 UI Framework Independent: Implement your own UI adapter for any frontend
• 📝 Strong Typing: Full TypeScript support for better development experience

Installation

npm install morph-agent
# or
yarn add morph-agent
# or
pnpm add morph-agent

Quick Start

Here's a simple example of creating a weather chatbot:

import { Agent, LLMSession, UiAdapter } from 'morph-agent';
import { CallApiStrategy } from 'llm-agent/strategies';

// 1. Implement LLMSession for your chosen LLM provider
class MyLLMSession implements LLMSession {
  async generate(prompt: string): Promise<string> {
    // Implement your LLM call here
    return 'LLM response';
  }
}

// 3. Create and configure your agent
const agent = new Agent({
  llm: new MyLLMSession(),
  capabilities: [CallApiStrategy],
  systemPrompt: 'You are a helpful assistant...',
  onEvent: (event) => console.log('Agent event:', event),
});

// 4. Start chatting!
await agent.chat('Hello!');

Core Concepts

LLM Session

The LLMSession interface is your connection to the Language Model:

interface LLMSession {
  generate(prompt: string): Promise<string>;
}

Example implementation for OpenAI:

import OpenAI from 'openai';

class OpenAILLMSession implements LLMSession {
  constructor(apiKey: string, model = 'gpt-4') {
    this.openai = new OpenAI({ apiKey });
    this.model = model;
  }

  async generate(prompt: string): Promise<string> {
    const completion = await this.openai.chat.completions.create({
      model: this.model,
      messages: [{ role: 'user', content: prompt }],
    });
    return completion.choices[0]?.message?.content || '';
  }
}

Capability Strategies

Capabilities are tools that your agent can use. Built-in strategies include:

• CallApiStrategy: Make HTTP requests
• More coming soon!

Create custom capabilities:

class MyCustomStrategy implements CapabilityStrategy {
  kind = 'myCustomAction';
  description = 'My custom action';
  signature = '{"param1": "string", "param2": "number"}';
  
  async execute(task: Task): Promise<TaskResult> {
    // Implement your custom capability
    return {
      id: task.id,
      status: 'success',
      output: 'Result',
    };
  }
}

Agent Events

Monitor your agent's actions through events:

agent.options.onEvent = (event: AgentEvent) => {
  switch (event.type) {
    case 'llmResponse':
      console.log('LLM said:', event.data.rawText);
      break;
    case 'taskStart':
      console.log('Starting task:', event.data.task.kind);
      break;
    case 'taskResult':
      console.log('Task result:', event.data.output);
      break;
    // ... handle other events
  }
};

Examples

Check out the /examples directory for complete implementations:

• cli/weather-chatbot.ts: A command-line weather information bot
• More examples coming soon!

Contributing

Contributions are welcome! Please read our Contributing Guide for details.

License

This project is licensed under the MIT License - see the LICENSE file for details.