The AI Config Generator creates complete YAML configurations from your natural language descriptions. No need to memorize YAML syntax or configuration options—just describe what you want your agent to do in any language.
What You’ll Learn This guide shows you how to:
Access the AI Config Generator in Atthene Agents Studio
Write descriptions that produce good results
Use generated configurations in your projects
How to Use the AI Config Generator
Open the Generator
In Atthene Agents Studio, click the ✨ Generate Config button in the top toolbar.
Describe Your Agent
In the modal that appears, describe what you want your agent to do. Be specific about:
What task the agent should perform
What data it needs to extract or analyze
What tools it should use (if any)
For example: Build a research assistant that can search the web for information about AI trends, analyze the findings, and create a comprehensive summary report.
Generate
Click Generate Config and wait a few seconds. The AI will create a complete YAML configuration.
Review and Test
The generated configuration appears in your YAML editor. Review it, then click Load Configuration to test it.
Writing Good Descriptions ✅ Good Examples Data Extraction Analyze customer feedback and extract sentiment, key topics, and actionable insights Customer Support Read customer support tickets, classify them as technical, billing, or general, and route each type to the appropriate specialist agent Content Review Review blog posts for quality. If quality is good, approve it. If quality is poor, send it back for revision. ❌ Avoid Too vague : “Create an AI agent” — What should it do?
Too long : Paragraph after paragraph of requirements — Keep it to 2-3 sentences
Implementation details : “Use GPT-4 with temperature 0.7” — Let the AI handle the technical details
Try These Examples Copy and paste these into the generator to see what it creates:
Example 1: Meeting Analyzer Analyze meeting transcripts and extract action items, decisions made, and key discussion points Example 2: Email Classifier Classify incoming emails as urgent, normal, or low priority based on content and sender Example 3: Quality Check Review code submissions and check if they meet our standards. If they pass, approve them. If not, send back with improvement suggestions. Tips for Best Results Be specific about data : If you need to extract specific information, list the exact fields you want (name, email, address, etc.)
Mention tools when needed : If your agent needs to search the web or access a knowledge base, say so in your description
Describe the workflow : If you need multiple steps (“first do X, then do Y”), describe them in order
Start simple : Generate a basic version first, test it, then add complexity
Common Issues “Not what I wanted” : Your description might be too vague. Try being more specific about what the agent should do and what data it needs.
“Too complex” : The generator created more than you need. Simplify your description to just the core task.
“Almost perfect” : You can manually edit the generated YAML to fine-tune it. The generator gives you a solid starting point.
What’s Next? Now that you can generate configurations:
Test Your Agent Learn how to test and run your generated agent
