Building Your First Agent
Creating a basic AI Agent in the AIGNE framework is intuitive and efficient. This guide will walk you through building and running your first intelligent Agent from scratch.
Basic Process#
Creating and using a basic Agent mainly involves the following key steps:
- Import Necessary Modules - Import framework core components and model interfaces
- Create AIGNE Instance - Configure framework runtime environment and underlying models
- Configure and Create Agent - Define Agent role and behavioral guidelines
- Use Agent to Process User Input - Invoke Agent and get response results
Let's understand the implementation details of each step:
Import Necessary Modules#
import { AIAgent, AIGNE } from "@aigne/core";
import { OpenAIChatModel } from "@aigne/openai";
Here we import three key components:
- AIAgent - Core class for creating and managing AI Agents, encapsulating Agent behavior and capabilities
- AIGNE - Main framework entry point, responsible for coordinating Agent workflows and lifecycle management
- OpenAIChatModel - Provides interface for interacting with OpenAI models, enabling Agents to access powerful LLM capabilities
Create AIGNE Instance#
const aigne = new AIGNE({
model: new OpenAIChatModel({
apiKey: process.env.OPENAI_API_KEY,
model: "gpt-4o-mini",
}),
});
Key points of this step:
- AIGNE instance serves as the core runtime environment, connecting Agents with underlying LLM models
- We chose OpenAI's "gpt-4o-mini" model as the Agent core, providing a good balance of performance and cost
- API key is securely read from environment variables, following development best practices
- Framework supports flexible switching between different provider models, such as Anthropic Claude, Google Gemini, etc.
Configure and Create Agent#
const agent = AIAgent.from({
instructions: "You are a helpful assistant for Crypto market cap",
inputKey: "message",
});
In this concise yet powerful configuration:
- Use
AIAgent.from()factory method to create Agent instance, simplifying the initialization process - Define Agent's professional domain and behavioral boundaries through the
instructionsparameter - This seemingly simple configuration is actually sufficient to create an intelligent assistant focused on cryptocurrency market cap information
- Framework automatically handles underlying prompt engineering and context management, allowing you to focus on business logic
Use Agent to Process User Input#
const result = await aigne.invoke(agent, { message: "What is crypto?" });
console.log(result);
// Output: { message: "Cryptocurrency, often referred to as crypto, is a type of digital or virtual currency that uses cryptography for security" }
This invocation process demonstrates:
- Use
aigne.invoke()method to pass user questions to Agent for processing - This method returns a Promise, requiring
awaitto wait for response completion - Input parameters include Agent instance and user's question text
- Response result uses standardized format, where
$messagefield contains Agent-generated answer
Example Code#
Below is a complete example showing how to create a basic Agent and use it to respond to user questions:
import { AIAgent, AIGNE } from "@aigne/core";
import { OpenAIChatModel } from "@aigne/openai";
const aigne = new AIGNE({
model: new OpenAIChatModel({
apiKey: process.env.OPENAI_API_KEY,
model: "gpt-4o-mini",
}),
});
const agent = AIAgent.from({
instructions: "You are a helpful assistant for Crypto market cap",
inputKey: "message",
});
const result = await aigne.invoke(agent, { message: "What is crypto?" });
console.log(result);
// Output: { message: "Cryptocurrency, often referred to as crypto, is a type of digital or virtual currency that uses cryptography for security" }
Tips#
- Precise Instruction Design: Agent performance and behavior largely depend on the
instructionsparameter; clear, specific instructions can significantly improve response quality - Model Selection Strategy: Different models have trade-offs in capabilities, speed, and cost; choose the most suitable model based on specific application scenarios
- Security and Privacy Protection:
- Always manage API keys through environment variables or secure services
- Implement appropriate access control and permission boundaries
- Validate and filter user input to prevent potential injection attacks
- Unified Response Handling: The AIGNE framework returns standardized response format
{ $message: "..." }, facilitating consistent processing in applications