Runs on Atlassian vs Customer-managed egress: this needs to change

Steffen, this matches the tradeoff I hit while building a small Forge-native Confluence app aimed at Runs on Atlassian. ROA is a sharp product constraint, not just a badge. In practice it means: - Forge-hosted compute + residency-enabled Forge storage (or host entity/content properties) - no remotes, no Connect modules, no dynamic web triggers - no “fetch whatever the page already embeds” unless you declare egress (and that usually kills ROA) So for a slideshow / renderer of Confluence content that must show Tableau, YouTube, or other embeds already on the page, CME (or wildcard client fetch)