Operator asked for a watering schedule as separate lines (bulleted, then numbered, then plain). Every format the Companion produced rendered as a single collapsed line in the chat bubble, regardless of how the Companion structured its output. The Companion itself correctly identified this as a display-side issue, not a content issue — it repeatedly confirmed its own output had genuine line breaks. This is a chat-message rendering bug (likely whitespace/newline handling in the message component), not a content or classifier problem.
In the same conversation, the Companion told the Operator "I'm not able to view images" — but this session has repeatedly confirmed working image processing elsewhere (vision descriptions, "Read with Describe what's in this image"), including on the very screenshot attached in this same turn, which was successfully processed ("Read with Describe what's in this image · Image · PNG"). The Companion's claim contradicts its own working capability in the same conversation — a capability-awareness or prompt issue, not a genuine missing feature.
No conflict. Both are bugs — a rendering defect and a capability-misrepresentation defect — not settled-decision questions.
No change to the image-processing pipeline itself (already working, per this session's repeated confirmation). No change to CR-2026-146's voice-capture work.