CloudThinker
  • Get Started
    • Welcome to CloudThinker
    • Key Features
    • Announcements
      • CloudThinker Beta Launch
  • Quick Start
  • HOW TO GUIDE
    • Workspace
      • Setup Your Workspace
    • Knowledge Base
      • CloudThinker Knowledge Base Guide
    • Agent Orchestration
      • Multi-agents orchestration
        • CloudThinker Prompting Guide
      • Agent
        • Build your own Agents
        • Tony - Database Engineer
        • Kai - Kubernetes Engineer
        • Alex - Cloud Engineer
        • Anna - General Manager
      • Connectors
        • MCP Manager
        • Model Context Protocol (MCP)
      • Memory
  • Task Management
    • Schedule Tasks
  • Management
    • Slack Integration
  • Management
    • User Roles & Permissions
    • Payment & Subscription
  • Learn More
    • Prompting Tips
    • Trouble Shooting
Powered by GitBook
On this page
  • πŸ“š Introduction: The Power of Effective Communication
  • πŸ” Understanding the Two Prompting Philosophies
  • πŸ“‹ Structured Task-Based Prompts: The Blueprint Approach
  • πŸ’­ Free-Style Conversational Prompts: The Coffee Chat Approach
  • 🌟 Prompt Crafting Mastery: Tips for Excellence
  • πŸšΆβ€β™€οΈ Getting Started: Your First Prompts
  • πŸ”„ The Feedback Loop: Refining Your Prompts
  • 🌈 Real-World Scenarios: Choosing Your Approach
  • ⭐ Conclusion: Your AI-Enhanced Cloud Journey
  1. HOW TO GUIDE
  2. Agent Orchestration
  3. Multi-agents orchestration

CloudThinker Prompting Guide

Transforms complex cloud operations into simple, natural language conversation.

πŸ“š Introduction: The Power of Effective Communication

Welcome to CloudThinker.IO, your AI-powered cloud operations partner! This guide will transform how you interact with our intelligent agents through the art of effective prompting.

Think of our AI agents as your cloud team members - always available, infinitely scalable, and ready to help. Just as with human assistant, how you communicate your needs dramatically influences the quality of results you receive.


πŸ” Understanding the Two Prompting Philosophies

CloudThinker offers two distinct yet complementary approaches to interact with our AI ecosystem:

πŸ—οΈ 1. Structured Task-Based Prompts

What it is: A formal, organized approach that clearly defines goals, provides step-by-step instructions, and establishes boundaries.

When to use it: For complex analytical tasks, multi-step processes, or when you need specific, formatted outputs.

πŸ’¬ 2. Conversational Free-Style Prompts

What it is: A natural, chat-like approach using @mentions and #action tags to direct your request to specific AI specialists.

When to use it: For quick insights, exploratory questions, or ongoing dialogues about your cloud environment.


πŸ“‹ Structured Task-Based Prompts: The Blueprint Approach

Format Template:

Goal: [One clear, measurable objective]

Instructions:
- [Step 1 of the process]
- [Step 2 of the process]
- [Data to analyze or transform]
- [Desired output format]

Constraints:
- [Boundaries of the analysis]
- [Exclusions to consider]
- [Time periods or other limits]

✨ Real-World Success Story:

The Situation: Sarah, Cloud Architect at TechCorp, faced quarterly budget pressure and needed to quickly identify optimization opportunities.

Her Prompt:

Goal: List all instances with <30% CPU utilization in last 30 days

Instructions:
- Pull CloudWatch metrics for all EC2 instances
- Calculate average CPU usage per instance
- Flag instances below 30% CPU utilization
- List potential savings by instance type
- Sort recommendations by cost impact

Constraints:
- Analyze only running instances
- Exclude instances tagged "production=true"
- Use last 30 days of data

The Result: CloudThinker identified 12 underutilized instances costing $4,200 monthly that could be downsized or converted to Spot instances, potentially saving $3,400. Sarah implemented the changes and became the department's cost-optimization hero.

🎯 Why This Works:

  1. Clear objective: The goal statement is specific and measurable

  2. Logical workflow: Instructions follow a natural analytical progression

  3. Defined boundaries: Constraints prevent unwanted recommendations

  4. Actionable output: Results are sorted by business impact


πŸ’­ Free-Style Conversational Prompts: The Coffee Chat Approach

Format Template:

@[agent-name] [natural language request] #[action-tag] [specific outcomes desired]

πŸ“± Available Agent Specialists:

  • @alex - Cloud infrastructure expert (AWS, Azure, GCP)

  • @kai - Kubernetes and containerization specialist

  • @anna - General manager for high-level strategy and coordination

  • @olivier - Security and compliance specialist

  • @tony - Database performance and optimization expert

🏷️ Common Action Tags:

  • #recommend - Suggest specific actions or alternatives

  • #alert - Highlight critical issues requiring attention

  • #chart - Display metrics or data visualizations for analysis

✨ Real-World Success Story:

The Situation: Miguel, DevOps Engineer at a growing startup, noticed performance complaints but couldn't identify the cause.

His Prompt:

@alex analyze T-type instances for CPU credit exhaustion. #alert when instances frequently run out of CPU credits and #recommend appropriate sizing.

The Result: Alex immediately identified three t3.micro instances regularly exhausting CPU credits during daily traffic peaks at 2-3PM. CloudThinker visualized the pattern, recommended upgrading two instances to t3.medium and converting one to a c5.large, with projected performance improvements and cost implications. Miguel implemented the changes and the performance complaints disappeared.

🎯 Why This Works:

  1. Direct specialist access: Engages the most appropriate AI expert

  2. Natural communication: Uses everyday language for accessibility

  3. Clear action requests: Tags guide specific types of responses

  4. Conversational efficiency: Gets straight to the problem without verbose formatting


🌟 Prompt Crafting Mastery: Tips for Excellence

For Structured Task Prompts:

  1. Start with the end in mind - Define your goal in measurable terms

  2. Break complex tasks into steps - Make each instruction clear and specific

  3. Provide context in constraints - Help the AI understand your environment

  4. Request specific formats - Ask for tables, charts, or prioritized lists when needed

  5. Iterate when necessary - Refine your prompt based on initial results

For Conversational Prompts:

  1. Choose the right specialist - Direct your question to the most appropriate expert

  2. Use multiple action tags - Combine #analyze with #recommend for comprehensive help

  3. Be conversationally specific - Casual tone works, but include necessary details

  4. Follow up naturally - Ask "Why do you recommend that?" or "Can you explain more about X?"

  5. Combine agent expertise - Try "@kai and @olivier review our cluster security configuration"


πŸšΆβ€β™€οΈ Getting Started: Your First Prompts

Try these starter prompts:

For Cost Optimization:

Goal: Identify cost reduction opportunities across my cloud environment

Instructions:
- Analyze last 90 days of billing data
- Identify underutilized resources and potential savings
- Look for usage patterns that suggest optimization opportunities
- Recommend specific actions ranked by ROI

Constraints:
- Consider both AWS and Azure environments
- Exclude development environments (tagged as "env=dev")
- Focus on opportunities saving at least $100/month

For Performance Investigation:

@tony examine our production RDS instances and #analyze query performance over the past week. #recommend any index changes or query optimizations that could improve response times.

For Security Assessment:

@olivier review our S3 bucket configurations. #alert me to any public access risks and #recommend best practices we should implement.

πŸ”„ The Feedback Loop: Refining Your Prompts

Remember that working with AI is collaborative. If your initial results aren't exactly what you need:

  1. Be more specific - "Could you focus specifically on our Lambda cold start issues?"

  2. Ask for alternatives - "What other approaches might work here?"

  3. Request explanations - "Why did you recommend this particular solution?"

  4. Adjust the format - "Could you present this as a prioritized action plan instead?"

The more you interact with CloudThinker, the more it learns your preferences and environment.


🌈 Real-World Scenarios: Choosing Your Approach

When you need to...
Best Approach
Example

Perform detailed cost analysis

Structured

Goal: Analyze serverless vs. container costs...

Get quick advice on an issue

Conversational

@kai why might our pods be crashlooping?

Generate comprehensive reports

Structured

Goal: Create monthly security compliance report...

Learn about best practices

Conversational

@olivier explain zero-trust model for our AWS environment

Troubleshoot urgent problems

Conversational

@alex our website is slow. #analyze load balancer metrics now

Plan infrastructure changes

Structured

Goal: Evaluate migration options from EC2 to ECS...


⭐ Conclusion: Your AI-Enhanced Cloud Journey

CloudThinker's AI agents are designed to be your trusted partners in cloud operations excellence. By mastering these prompting techniques, you're unlocking their full potential to transform your operations.

Remember that CloudThinker is always learning. Your interactions help our agents better understand your unique environment and challenges, becoming more valuable with every conversation.

Need help with prompting? Simply type:

@anna I'm not sure how to ask for what I need. Can you help me craft an effective prompt for [your situation]?

Our General Manager agent will guide you through creating the perfect prompt for your specific needs.


CloudThinker.IO - From Cloud Complexity to Operational Clarity

PreviousMulti-agents orchestrationNextAgent

Last updated 13 days ago