add-image-vision

You enable image understanding by merging this skill into your ClaudeClaw setup, rebuilding the container, and restarting the service. This allows agents to download, resize with sharp, and process WhatsApp image attachments as base64 multimodal content blocks.

Yes, sharp is used to resize WhatsApp images before converting them to base64. This ensures the image attachments are properly formatted as multimodal content blocks for the ClaudeClaw agent to interpret within the conversation context.

The workflow for adding base64 image vision involves a phase-driven process: pre-flight validation, code changes, container rebuild, and integration with the agent-runner. This ensures the agent successfully processes base64-encoded multimodal content blocks.

Yes, adding image vision requires existing skill scaffolding within ClaudeClaw. The implementation depends on the sharp library for image processing and integrates with the existing agent-runner workflow to pass base64-encoded image blocks.

Your ClaudeClaw agent needs image vision to interpret visual content shared in WhatsApp group chats. By downloading and encoding attachments as base64 multimodal blocks, the agent gains situational awareness of images sent within ongoing conversations.

A limitation is that processing WhatsApp images requires existing skill scaffolding and sharp dependency. The workflow targets WhatsApp groups where image content needs interpretation, saving images to the group workspace for agent context.