If you have an idea for a custom visualization in Oracle Analytics Cloud (OAC), an AI coding assistant can help turn it into an uploadable extension. You can follow this approach without installing a local SDK.
Want to see it in action? Watch this short demo video to see how an AI coding assistant can help create a custom extension for OAC.
Start by giving the AI assistant one or more custom visualization ZIP files that you have already confirmed work in OAC. These are reference packages: the assistant can adapt their files for your chart while preserving the package structure OAC expects. You can use your own extensions or examples from the Oracle Analytics Extensions Library and community posts, after confirming they work in your environment.
A practical workflow
1. Define the visualization. Describe the chart you want and how it should behave in an OAC workbook. Be clear about the Grammar Panel: which roles accept attributes, which accept measures, whether any accept both, and which roles are required or optional. Also describe any aggregation, formatting, tooltips, legends, or interactions that matter to your design. The Grammar Panel, data mapping, and data source integration articles in our custom visualization series provide useful context for these decisions.
2. Build from the working package. Give your requirements and reference ZIPs to an AI coding assistant. Have it adapt the extension, connect the chart to live workbook data, and prepare a new ZIP. As a good practice, ask it to check that the package is complete, its files point to the right resources, and its version is clear before you upload it. This catches avoidable packaging issues early.
3. Test in OAC and iterate. Upload the ZIP through Console > Extensions, add the visualization to a workbook, and see how it performs. If something fails, give the AI assistant the exact error message or a clear description of what you observed. Have it correct the package, rebuild it, and test the new version in OAC.
Expect a few cycles. A ZIP may have an unrecognized folder structure, or a chart may load without showing the expected data. There could be package structure errors. Sharing those errors with the AI assistant will help correct the packages and reach working versions.
If an uploaded extension does not appear under Custom Visualizations, or OAC continues to show a previous version after an upload, try a hard refresh or Empty Cache and Hard Reload in your browser, then check again.
It is recommended to give each tested ZIP a distinct version so you can track which build produced each result.
The optional OAC Plugin Skill
Here’s a sample OAC Plugin Skill as a ZIP that one could use as a quicker starting point. It gives an AI coding assistant a reusable guide for gathering chart requirements, starting from a working extension, handling live data and common chart features, and checking the package before delivery. It can save you from repeating the same guidance for every visualization. The skill is optional and serves as an example; your own working ZIPs and clear requirements are enough to begin.
Here’s an example of using the skill to create a 100% stacked bar chart. It prompts to fill in a short request form. Following are the sample details provided to the skill:
Goal | Details Required |
|---|
Chart family | 100% stacked bar chart |
Category (attribute) | Product Category |
Value (measure) | Sales |
Color (optional attribute) | Customer Segment |
Aggregation | Sum of Sales |
The form also allows optional choices for sorting, number format, and other settings. Following are examples of extensions created using the above mentioned process. While these visualizations are available OOB in OAC, it is only meant to serve as an example of how to use the extension to quickly create your own visualization.
Once an extension works in OAC, the JavaScript and supporting files are available in its ZIP. You can inspect and modify them to keep improving the visualization.
The process is straightforward: start with a working OAC package, describe the visualization and its data roles, build with AI, then test and refine it in OAC. Errors are part of that cycle but the overall approach is efficient and saves time.
Download the optional OAC Plugin Skill:
Call to Action
- Use AI Vibe Coding to Build OAC Custom Extensions
- OAC Extensions Library
- Unlocking Custom Visualizations: Configuring the Grammar Panel
- Unlocking Custom Visualization Blog Series
- Custom Extension Community Post
