Skip to main content

​
Quick Reminder: Opening Monk

Click the Monk icon in your IDE sidebar or press Cmd+Shift+M (Mac) / Ctrl+Shift+M (Windows/Linux). The chat window appears - you’re ready to prompt.

​
Talk Naturally

Monk understands natural language. No commands to memorize, no syntax to learn. Just say what you want: 
deploy this project
show me the logs
how much does this cost?
scale the API server up
Deploy project with monk Not this: 
monk deploy --target=production --provider=aws
GET /api/logs?service=api&lines=100

​
Monk Interprets Your Intent

You describe what you want, not how to do it. You say: 
deploy this to the cloud
Monk figures out:
  • Analyze the code
  • Create containers
  • Provision infrastructure
  • Deploy services
  • Wire everything together
All the steps are autonomous. You don’t specify them.

​
Vague Prompts Work

Don’t know all the details? Start vague: 
You: Deploy this application

Monk: I see a React frontend and Node.js API.
      Which cloud provider would you like?
      - AWS
      - GCP
      - Azure
      - DigitalOcean

You: AWS

Monk: Which region?

You: us-east-1

Monk: [Proceeds with deployment]
Or be specific upfront: 
You: Deploy to AWS us-east-1, use RDS for the database

Monk: [Proceeds without questions]
More detail = fewer questions.

​
Monk Works Autonomously

Unlike chatbots that ask permission for everything, Monk executes autonomously. Monk doesn’t ask:
  • “Should I analyze your package.json?”
  • “Should I create a Dockerfile?”
  • “Should I configure networking?”
Monk does ask:
  • Credentials - When it needs cloud provider or API credentials
  • Important choices - Deployment options, cloud provider selection
  • Destructive actions - Confirmation before deleting resources
  • Clarification - When requirements are ambiguous

​
The Build → Deploy → Adjust Cycle

Monk follows a natural workflow:

​
1. Build

You: deploy this project
  • Monk analyzes your code
  • Creates containers
  • Generates configuration
  • Shows you what will be deployed

​
2. Deploy

You: [Approve after reviewing]
  • Monk provisions infrastructure
  • Deploys your application
  • Verifies everything works

​
3. Adjust

You: show me the logs
     scale up
     add a Redis cache
  • Monitor and modify as needed
  • Changes applied with zero downtime
Then repeat: Make code changes, deploy updates, adjust again.

​
Answering Monk’s Questions

When Monk asks, answer naturally: Direct answer: 
Monk: Which cloud provider?
You: AWS
Ask Monk to decide: 
Monk: Which cloud provider?
You: What do you recommend?
Monk: [Provides recommendation based on your needs]
Provide context: 
Monk: How should I deploy the database?
You: We need high availability, budget isn't a concern
Monk: I recommend AWS RDS Multi-AZ for high availability

​
Secure Credentials

When Monk needs credentials, you’ll see secure input prompts:
  • Values masked as you type
  • Never exposed to the AI
  • Stored encrypted on your infrastructure
  • Never sent to Monk servers
You provide credentials once per provider:
  • First time: Monk asks
  • Next time: Uses saved credentials

​
Common Patterns

Exploratory (understand before acting): 
analyze this project
what will it cost?
show me the configuration
Action (make something happen): 
deploy this
set up CI/CD
scale up
migrate to DigitalOcean
Information (check status): 
what's running?
show me the logs
how much am I spending?
Troubleshooting (fix issues): 
why is the API slow?
users are reporting errors
check the database performance

​
Next: Prompting Cookbook

The rest of this section is a cookbook - specific tasks with example prompts. Pick what you want to do: