2.3 KiB

Raw Permalink Blame History

Creating an AI Agent

Learn how to create an Mastra agent that can be used within your workflows for more intelligent content processing.

Creating a Content Analysis Agent

Create a new file for your agent in the src/mastra/agents directory. Use content-agent.ts as the name of the file with the following contents:

// src/mastra/agents/content-agent.ts
import { openai } from "@ai-sdk/openai";
import { Agent } from "@mastra/core/agent";

export const contentAgent = new Agent({
  name: "Content Agent",
  description: "AI agent for analyzing and improving content",
  instructions: `
    You are a professional content analyst. Your role is to:
    1. Analyze content for clarity and engagement
    2. Identify the main themes and topics
    3. Provide a quality score from 1-10
    4. Suggest specific improvements
    
    Always provide constructive, actionable feedback.
  `,
  model: openai("gpt-4o-mini"),
});

Understanding the Agent

Name: Unique identifier for the agent
Description: What the agent does
Instructions: Detailed prompts that guide the AI's behavior
Model: Which AI model to use (GPT-4o-mini is fast and cost-effective)

Registering and Testing Your Agent

Open your src/mastra/index.ts file and add your agent (you may need to append it to the agents object in the Mastra class):

// Import your workflow
import { contentAgent } from "./agents/content-agent";

export const mastra = new Mastra({
  // Register your agent here
  agents: {
    contentAgent,
  },
  // ...Existing code
});

You can test this agent in the Playground by navigating to the Agents tab and selecting content-agent. Use the chat interface to validate the agent is working.

The agent should provide analysis of the content, including themes, quality assessment, and improvement suggestions.

Why Use Agents in Workflows?

Agents add intelligence to workflows by:

Understanding context: AI can interpret meaning, not just process data
Generating insights: Provide analysis that simple logic cannot
Adapting responses: Give different feedback based on content type
Natural language output: Communicate results in human-readable form

Your AI agent is ready! Next, you'll learn how to integrate it into a workflow step.

2.3 KiB Raw Permalink Blame History