Skip to main content

Overview

Working with the Stardeck agent is as simple as having a conversation. There are no special commands to memorize, no technical syntax to learn—just describe what you want in plain language and the agent handles the rest. This guide covers practical tips to help you get the most out of your conversations with the agent.

Just Talk Naturally

The agent understands natural language. Instead of trying to sound technical, just explain what you want the way you’d tell a colleague. Natural prompts work best:
  • “I need a contact form that sends me an email when someone submits it”
  • “Make the header stick to the top when I scroll”
  • “Add a photo gallery where I can upload images”
  • “The button should be bigger and more noticeable”
You don’t need to say things like:
  • “Implement a POST endpoint with form validation and SMTP integration”
  • “Add position: sticky to the header component with a z-index”
  • “Create a CRUD interface for image assets with S3 storage”
The agent translates your intent into the right technical implementation. If something isn’t clear, it asks follow-up questions to make sure it understands exactly what you need.

When In Doubt, Just Ask

One of the most powerful things about the agent is that you can ask it anything. If you’re unsure about something, ask. Not sure what to do next?
  • “What should I focus on next?”
  • “What’s missing from my website?”
  • “What would make this page better?”
Not sure if something is possible?
  • “Can I let users log in with their Google account?”
  • “Is it possible to send automatic emails when someone signs up?”
  • “Can I accept payments on this site?”
Don’t know how to accomplish something?
  • “I want customers to book appointments online, how do we do that?”
  • “I need to show different prices for different countries”
  • “How can I let my team edit content on the site?”
The agent can guide you through unfamiliar territory. It knows what the platform can do and will suggest approaches you might not have thought of.

Uploading Assets

You can upload files directly in the chat to share them with the agent. This is useful for:
  • Logos and brand images — Your company logo, icons, brand assets
  • Photos — Product images, team photos, background images
  • Documents — PDFs, spreadsheets, content files
  • Design files — Screenshots of designs you want to replicate
  • Reference images for generation — Images to guide AI image generation (style transfer, variations, editing)
Upload your brand assets early in the project. Once the agent has your logo and brand colors, it can incorporate them into every design automatically.
You can also use uploaded images as references for AI image generation. Upload an image and ask the agent to modify it, create variations, or use it as style inspiration. Learn more about image generation.

Supported File Types

The agent accepts most common file types including images (PNG, JPG, SVG, WebP), documents (PDF, CSV, TXT), and more. If you’re unsure whether a file type is supported, just try uploading it.

The Agent Can See Images

The Stardeck agent has vision capabilities—it can see and understand images you share. This opens up powerful ways to communicate: Share screenshots for feedback:
  • “Here’s a screenshot of the bug I’m seeing”
  • “This is how it looks on my phone—can we fix the spacing?”
Show designs you want built:
  • “Build this design” (with an attached mockup)
  • “Here’s a wireframe of what I’m thinking”
Share examples for inspiration:
  • “I like how this website does their navigation” (with a screenshot)
  • “Can you make something similar to this?” (with an example image)
Visual references eliminate ambiguity. Instead of trying to describe what you want in words, you can show it.

Loading Data into Your App

If you have existing data—product catalogs, customer lists, content from another system—you can upload it and the agent will help import it into your application.

How It Works

  1. Upload your data file — CSV files work great, but the agent can also parse spreadsheets and other formats
  2. Describe what the data represents — “These are my products” or “This is a list of team members”
  3. The agent imports it — It parses the file, maps the columns to your database, and loads the data

Common Use Cases

  • Product catalogs — Import your inventory from a spreadsheet
  • Customer lists — Migrate contacts from another system
  • Content migration — Move blog posts, FAQs, or other content
  • Menu items — Load restaurant menus, service offerings, or price lists

Example Workflow

1

Prepare your data

Export your data as a CSV or gather it in a spreadsheet. Make sure column headers clearly describe what each column contains.
2

Upload and explain

Drag the file into the chat and tell the agent what it is: “Here’s my product catalog. Each row is a product with name, price, description, and category.”
3

Review and confirm

The agent shows you how it plans to import the data. Confirm it looks right, or provide corrections.
4

Import complete

Your data is now in your application, ready to display to users.
For large data imports, it’s helpful to start with a small sample to verify everything works correctly before importing the full dataset.

Tips for Better Results

Be Specific About What You Want

The more detail you provide, the better the result. Compare:
VagueSpecific
”Add a form""Add a contact form with name, email, and message fields that sends me an email"
"Make it look better""Use more white space and make the headings bigger"
"Add a menu""Add a navigation menu with links to Home, About, Services, and Contact”

Share Visual References

When you have a specific look in mind, sharing an image is worth a thousand words. Screenshots of websites you like, design mockups, or even rough sketches all help the agent understand your vision.

It’s a Conversation, Not a Command

Don’t worry about getting everything perfect in one message. You can:
  • Start with a rough idea and refine it
  • Ask the agent to show you options
  • Change your mind after seeing the result
  • Build incrementally, one feature at a time

Answer the Agent’s Questions Thoroughly

When the agent asks for clarification, detailed answers lead to better results. If it asks “What information do you want to collect from users?”, listing out each field specifically helps more than saying “the usual stuff.”