Current situation:

Rovo processes images only when they're uploaded into the Rovo chat directly. Rovo is unable to process images that are part of Confluence pages or Jira issue attachments.

Suggestion:

It would be great if Rovo chat and agents could process and answer questions based on the images uploaded to Confluence pages and Jira issues as well.

This limitation also applies to the Atlassian Rovo MCP Server, which returns attachment metadata (filename, size, URL) but can't retrieve the actual file content, because the authenticated session isn't passed to the attachment download endpoint. Customers using MCP clients (e.g., ChatGPT, Claude) are unable to read or process Jira attachment contents.