User Problem

Rovo Remix in Confluence Cloud allows users to transform native content (tables, lists, text) into visual formats like charts, infographics, and diagrams. However, Remix does not recognize or process the data output of custom macros built on the Forge framework.

This is because Remix reads the page's Atlassian Document Format (ADF), and Forge macros only store their directive (extension key + parameters) in ADF — not the rendered output. The actual content is rendered at view time in the browser (front-end only for UI Kit and Custom UI macros), so the AI layer cannot "see" what the macro displays.

For enterprise customers and Forge app developers who rely on custom macros to present specialized data (dashboards, reports, filtered views), this means Remix is unusable on their most important content.

Suggested Solutions

1. Manifest opt-in property: Provide an API or manifest property in Forge (e.g. extending adfExport or a new agenticExport) that allows developers to opt-in to Remix capabilities and declare how their macro's data should be exposed to AI tools.

2. ADF/data summary parsing: Allow Remix to parse the exported ADF or a data summary provided by a Forge macro to generate AI-driven visualizations — even if the full rendered view isn't available.

3. Remix toolbar accessibility: Ensure the Remix sparkle icon/toolbar is accessible for custom macro blocks, so users can trigger Remix on Forge macro output the same way they do on native tables.

Current Workarounds

• Developers can declare an adfExport function in their Forge manifest to make macro content backend-renderable — this doesn't enable Remix directly but positions the macro for future AI integration.
• Developers can use Forge Rovo Actions to define custom agent actions that let Rovo interact with their app's data programmatically (alternative path to AI transformation).
• Developers can use the Forge LLMs API to integrate Atlassian-hosted LLMs directly into their macro for native AI summarization.
• Users can manually export Forge macro data as a native Confluence table on the page, then Remix that table — but this defeats the purpose of the custom macro.