Agent Mode enables DinoAI to directly create and modify files in your project, leveraging warehouse metadata and context to automate analytics engineering tasks.
Agent Mode lets DinoAI create and modify files directly in your project. It's designed to turn natural language instructions into working dbt™ code.
Source definitions: Create or update sources.yml with warehouse metadata
Model creation: Generate staging, intermediate, and mart models
Documentation: Document multiple models in seconds
Project configuration: Configure dbt_project.yml based on your folder structure
External integrations: Pull information from Jira tickets or search web resources with Perplexity
Code standardization: Apply consistent formatting and patterns across your codebase
Click the DinoAI icon (🪄) in the right panel of the Code IDE
Agent Mode is the default - switch only if you're currently in Ask Mode
Type a prompt describing what you want to create or modify
Review the suggested changes in preview boxes
Click Accept to apply or Reject to decline
Adding Context Makes DinoAI Smarter
Provide context for better results:
"Add File" for a specific reference file
"Add Open Files" for all open editor tabs
"Add Directory" for entire folders
Learn more about context
DinoAI creates or modifies files based on your prompts, always showing previews before applying changes.
Connects to your data warehouse to understand your metadata, including: tables, columns, and relationships.
Built-in Safeguards
All changes require your explicit approval
DinoAI won't run expensive queries without permission
Follows your .dinorules configurations for consistency
The more specific your prompt, the better the results:
Want to see Agent Mode in action? Check out our detailed Use Cases.
Last updated
Was this helpful?
Was this helpful?
"Create a staging model for the customers table in my raw_data source
Update sources.yml to include the new marketing tables.
Document all models in the marts/finance folder.