Visual execution layer for DeepSeek Harness

Key features

• The mode is explicit and reversible
• CreateGenerate from natural language, use up to 14 image references, upload, paste, drag, or add an existing chat image.
• An ambiguous request such as “design a hiking backpack” is not sent directly to an image model.
• The selected image determines which tools appear.
• an empty canvas does not interrupt the conversation.