AsyncGraph

AsyncGraph is an advanced multi-agent development framework built with TypeScript, designed to create, deploy, and manage autonomous AI agents. It offers a flexible and extensible platform for developing intelligent agents that operate independently, adapt to dynamic environments, and optimize their performance in real-time.

About the Project

AsyncGraph empowers developers with:

• Modular Architecture: Easy integration and customization for specific use cases
• Reasoning and Response Capability: Transparent decision-making process
• Knowledge and Learning Capabilities: Continuous learning from interactions
• Economic Efficiency: 10x-15x cost reduction compared to other frameworks
• Transaction Execution Support: Handles both on-chain and off-chain transactions

Why TypeScript?

TypeScript provides:

• Robust type safety
• Extensive library support
• Seamless integration capabilities
• Strong developer tooling
• Excellent scalability for complex applications

Products You Can Build

AsyncGraph enables the development of various autonomous agents:

DeFi Applications

• Autonomous trading bots
• Yield optimizers
• Arbitrage agents
• Swapping agents
• Bridging agents for cross-chain operations

Security & Management

• Security monitoring agents
• Contract auditing agents
• Automated contract management tools
• Real-time threat detection systems

Community & Gaming

• Community engagement bots
• Gaming AI agents
• Interactive gameplay assistants
• Social AI moderators

Betting & Analytics

• Decentralized betting agents
• Market analysis tools
• Risk assessment systems

Prerequisites

• Node.js (v18 or higher)
• TypeScript (v4.5 or higher)

Installation

# Install using npm
npm install

# Or using yarn
yarn

Environment Variables

Create a .env file in the root directory with the following configurations:

OPENAI_API_KEY=
ANTHROPIC_API_KEY=
MONGODB_URI=
DB_NAME=

Start the Application

Development Mode

npm run dev

Production Mode

npm run start

Integrating New Tools

AsyncGraph provides a flexible plugin system for integrating new tools:

1. Create Tool

Create a new file in the tools folder with the following structure:

import { tool } from "@langchain/core/tools";

export const CustomTool = tool(
  async ({ query }: { query: string }) => {
    // Tool implementation
  },
  {
    name: "custom-tool",
    description: "Description of your tool",
  }
);

2. Export Tool

Export the tool from the file:

export the tool in the tools folder in the index.ts file

Tool Integration Guidelines

• Ensure your tool implements the required interfaces
• Include comprehensive error handling
• Add appropriate type definitions
• Document tool capabilities and parameters
• Include usage examples

CURL Example For Testing

New Query

curl -X POST \
  http://localhost:3000/api/agent/query \
  -H 'Content-Type: application/json' \
  -d '{
    "query": "what is the weather in dubai"
  }'

Response

{
  {"success":true,"data":"The current weather in Dubai is clear with a temperature of 26.3°C (79.3°F). The wind is coming from the WNW at a speed of 7.6 kph (4.7 mph). The humidity is at 45%.","threadId":"thread_c4fa6585-397d-46c0-ae04-cb4da4102dd7"}
}

Continue The Conversation

curl -X POST \
  http://localhost:3000/api/agent/query \
  -H 'Content-Type: application/json' \
  -d '{
    "query": "what about india",
"threadId": "thread_c4fa6585-397d-46c0-ae04-cb4da4102dd7"
  }'

Response

{"success":true,"data":"The current weather in New Delhi, India is misty with a temperature of 21.2°C (70.2°F). The wind is coming from the West at a speed of 4.3 kph (2.7 mph). The humidity is at 46%.","threadId":"thread_c4fa6585-397d-46c0-ae04-cb4da4102dd7"}

Contributing

We welcome contributions! Please see our Contributing Guidelines for details.

License

License details here

Support

For support, please create an issue