Hi QuickSight Community,
I’m trying to fully automate QuickSight dashboard creation using Cursor IDE and an MCP server interacting with the QuickSight API.
While the concept is great, I’m struggling to get the LLM to generate valid and structurally sound Dashboard Definition JSONs. I consistently face two main issues:
1. API Validation Errors: The AI often misses required variables, properties, or fields, causing the QuickSight API to reject the JSON.
2. Broken Layouts: Even when the JSON passes validation, the generated visuals overlap, are obscured, or lack proper dimension/coordinate configurations, making them unreadable (especially in Free-form layouts).
My questions:
-
Has anyone successfully implemented a workflow to generate full QuickSight dashboards using AI?
-
Are there specific system prompts, schemas, or best practices you use to ensure valid JSON definitions?
-
How do you guide the LLM to calculate dynamic layouts (positions & sizes) correctly?
Any advice, workarounds, or examples would be hugely appreciated!
Thanks!
Hi @chanhlm,
I do not believe there is any documentation for the use case of generating dashboards in a workflow, but I would definitely to follow the JSON dashboard schema (example command mentioned here) and define your dashboard based on that. Feel free as well to look into other current AWS SDK documentation.
In terms of styling the dynamic layouts, try specifying what visual styling options you want to adjust and see if the workflow is able to understand and execute consistently.
Please feel free to let us know if you are able to make good progress on your use case and let us know if you have any other relevant questions!
Great timing! We just launched generative dashboard creation out-of-the-box. From the Analysis page, you will see a new option to “generate analysis”. We’d love to hear what you think
Here is the doc: Generating an analysis with natural language prompts - Amazon Quick
I am currently on an Author Pro account, but I haven’t seen this feature available on my end yet. That sounds fantastic, though—it seems highly useful, and I will definitely test it out as soon as I can!
That being said, my ultimate goal is to leverage external AI agents alongside the QuickSight MCP to achieve a deeper level of automation via Cursor. I want to go beyond just generating analyses and extend this to automating dataset creation, writing SQL queries, and more.
Thank you so much for sharing this information!
What region are you in?
Thanks for sharing your end to end use case. Have you tried using flows or automations?
Hi @chanhlm,
Hope everything is well! Just wanted to check back in and confirm if you saw alaresch’s most recent reply. Please help the community by marking this answer as “Solution” or following up in general within the next 3 business days!
Thank you!
Hi @chanhlm,
Since I haven’t received any further updates from you, I’ll treat this inquiry as complete for now. If you have any additional questions, feel free to create a new post in the community and link this discussion for context.
Thank you!