Introducing the DinoAI Prompt Library

Since launching DinoAI "Cursor for Data" with agent mode, we've watched hundreds of analytics engineers discover what AI-powered data development looks like. The traction has been intense—and it's taught us something important.

While software engineers have been crafting prompts and leveraging AI for years, analytics engineers are just getting started. The data tooling ecosystem is just now starting to embrace AI the way software development already has, which means most data practitioners are flying blind when it comes to prompt engineering.

That changes today 🚀.

The Problem: Analytics Engineers Need Better Prompts

Here's what we learned from talking to our users: analytics engineers face repeatable problems every day, but they struggle to articulate these problems in ways that LLMs can understand and solve effectively.

You know the drill. You spend 20 minutes crafting the perfect prompt to generate code for your intermediate models, getting the CTEs formatted just right, only to save it in a random text file that you'll inevitably lose. Then you spend another 20 minutes recreating it next week.

Meanwhile, your team is solving similar problems with completely different approaches, reinventing the wheel instead of sharing what works.

Introducing the DinoAI Prompt Library

The DinoAI Prompt Library solves this. It's the first prompt library built specifically for analytics engineers, and it's available today within the Paradime platform.

Here's how it works:

Ready-Made Prompts: We've identified the most common analytics engineering tasks and created battle-tested prompts that deliver consistent results. This covers 70-80% of what most analytics engineers need AI help with.

Customizable and Extensible: The entire prompt library lives in a .dinoprompts file—a simple YAML text file you can edit directly. Add your own prompts, modify existing ones, and tailor the AI experience to your specific needs.

prompts:
- name: 'Generate pull request description'
  prompt: |
    Write a pull request description for the following code changes using markdown:
    {{ git.diff.withOriginDefaultBranch }}
    Include the following sections:
    - Summary of changes
    - Motivation for the changes
    - How to test the changes
    
- name: 'Generate a dbt sources file'
  prompt: |
    Generate a complete dbt sources.yml file following this step-by-step process:
    1. LIST ALL AVAILABLE DATABASES for me to select from (Skip if I specify: [DATABASE])
    2. After database selection, LIST ALL SCHEMAS in that database (Skip if I specify: [DATABASE].[SCHEMA])
    3. Configure ALL sources with standardized freshness checks using database metadata:
       - Warning threshold: 24 hours
       - Error threshold: 48 hours
       - No loaded_at_field (use database metadata)
    4. Include ALL tables in the selected schema(s) with:
       - Clear business descriptions for each table
       - Appropriate tags based on table naming patterns or content
       - Do NOT list individual columns
    5. Include example source() usage

Here is a small excerpt, where you just define a name and a multi-line prompt and we take care of the rest. We have also introduced useful substitutions like {{ git.diff.withOriginDefaultBranch }} and we are adding more as we go along.

Team Sharing: Found a prompt that transforms your workflow? Share your .dinoprompts file with your team. Now everyone benefits from your research, and your organization becomes genuinely AI-first.

Quick Access: Use the prompt quick-open feature to instantly access any prompt in your library. No more hunting through text files or trying to remember that perfect prompt from last month.

We have also compiled a set of best practices on how to create effective prompts for data work as follows:

# DinoAI Prompts Collection
# --------------------------------
# This file contains a collection of prompts designed to assist with your development workflows.
# These prompts are available in DinoAI by using the "[" hotkey or clicking on the "prompts" button in the DinAI input
#
# PRO TIPS:
# * You can update or create new prompt templates using the same format to customize available prompts in DinoAI
# * Use DinoAI to help you write new templates based on your own use case. For example: 
# * Tell DinoAI "Can you add a prompt template in my .dinoprompts to help me migrate a dbt model to incremental"
# * You can built in variables in your prompts to access context on the fly
# * Check our docs and tutorials here -> https://docs.paradime.io/app-help/documentation/dino-ai/dino-prompts
#
# How to use:
# 1. Select the appropriate prompt based on your current task
# 2. Provide any required information (file paths, database details, etc.)
# 3. The assistant will guide you through the process or generate the requested content
#
# Included prompt categories:
# - Pull request documentation
# - Source configuration management
# - Model generation and refactoring
# - Documentation and testing
# - Jira/Linear Issue helper

Why This Matters

A prompt library does more than organize your AI interactions—it accelerates your entire development workflow. Instead of starting from scratch every time, you build on proven patterns that actually work.

This is bigger than individual productivity. When teams share prompt libraries, they create organizational knowledge that makes everyone more effective with AI. Platform teams can provide standardized prompts to dbt contributors across the organization, ensuring consistency and best practices.

We're the only Code IDE offering this capability. No other IDE provides prompt library functionality at the moment. We're pioneering this because we believe analytics engineers deserve the same AI-powered productivity that software engineers have been enjoying. The format is simple, easy to use and easy to share.

What's Next

This is just the beginning. We'll be releasing more prompts as we learn how teams use the library, and we want your input on what to build next.

Our goal remains the same: provide the best "Cursor for data" experience possible. The prompt library is another step toward eliminating 80-90% of the operational overhead that slows down data teams.

The DinoAI Prompt Library is releasing today for all Paradime users. Start building your prompt collection today, and let us know what you create with it.