Plan-Execute Agent Example

A Plan-Execute agent that creates a plan first, then executes each step systematically.

Overview

The Plan-Execute pattern:

1. Plan - Create a detailed plan to solve the task
2. Execute - Execute each step sequentially
3. Verify - Check if the goal is achieved
4. Replan - Adjust plan if needed

Complete Example

import { createPlanExecuteAgent } from '@agentforge/patterns';
import { ChatOpenAI } from '@langchain/openai';
import {
  webScraper,
  fileWriter,
  jsonParser
} from '@agentforge/tools';

const agent = createPlanExecuteAgent({
  model: new ChatOpenAI({ model: 'gpt-4' }),

  tools: [webScraper, fileWriter, jsonParser],

  plannerPrompt: `Create a detailed step-by-step plan to accomplish the task.
Each step should be clear and actionable.
Consider dependencies between steps.`,

  executorPrompt: `Execute the current step using available tools.
Be thorough and accurate.
Report results clearly.`,

  maxSteps: 10
});

// Use the agent
const result = await agent.invoke({
  messages: [{
    role: 'user',
    content: 'Research the top 5 AI frameworks and create a comparison report'
  }]
});

console.log('Final Report:', result.messages[result.messages.length - 1].content);

Output Example

📋 Plan Created:
1. Search for top AI frameworks
2. Gather information about each framework
3. Compare features, pros, and cons
4. Create structured comparison
5. Write report to file

⚙️ Executing Step 1: Search for top AI frameworks
✅ Found: TensorFlow, PyTorch, LangChain, Hugging Face, JAX

⚙️ Executing Step 2: Gather information about each framework
✅ Collected detailed information

⚙️ Executing Step 3: Compare features
✅ Comparison complete

⚙️ Executing Step 4: Create structured comparison
✅ JSON structure created

⚙️ Executing Step 5: Write report
✅ Report saved to ai-frameworks-comparison.md

📄 Final Report: Successfully created comparison report...

When to Use Plan-Execute

Best for:

• Research tasks
• Multi-step workflows
• Report generation
• Data analysis
• Complex problem-solving

Key Features

• ✅ Structured Planning - Clear step-by-step approach
• ✅ Sequential Execution - One step at a time
• ✅ Progress Tracking - Monitor execution
• ✅ Replanning - Adjust if needed
• ✅ Verification - Check goal achievement

Next Steps

• Reflection Pattern - Add self-improvement
• Multi-Agent System - Coordinate multiple agents