Current behavior:
Rovo Agents can read images when they are attached directly to a Jira issue as standalone image files (JPG, PNG, etc.).
However, when images are embedded inside file attachments (e.g., an XLSX with 3 embedded images, a PDF with screenshots, a DOCX with diagrams), Rovo cannot extract or interpret that visual content.
Expected behavior:
Rovo Agents should be able to extract and interpret images embedded inside file attachments, using the same vision capabilities already available for standalone image attachments.
Why this matters:
In real workflows, users attach documents (XLSX reports, PDF specs, DOCX requirements) that contain images as the primary content. The agent receives no information from these files, blocking AI-assisted analysis entirely.
Evidence:
Tested with an XLSX file containing 3 embedded images — openpyxl confirms the images exist (ws._images = 3) but Rovo returns no visual content from the file.
