add-image-vision

To add image understanding to WhatsApp agents, this skill downloads and resizes image attachments, then passes them to Claude as base64-encoded multimodal content blocks for visual interpretation.

Yes, you can process WhatsApp images in both group chats and self-chat scenarios, as the skill handles caption parsing and robust image processing across these different messaging contexts.

Yes, you need the sharp dependency installed, because the skill uses it to resize downloaded WhatsApp image attachments before packaging them into multimodal content blocks for Claude.

Multimodal vision with Claude works by encoding resized image attachments as base64 content blocks, allowing the model to process the visual data alongside text captions from WhatsApp messages.

The best way to send WhatsApp images to a Claude agent is using an integrated pipeline that downloads the attachment, resizes it with sharp, and delivers it as a multimodal content block.