Summary

Add image upload support (clipboard paste, drag-and-drop, file picker) to Control UI / WebChat for vision-capable models.

Problem to solve

Control UI / WebChat does not support image uploads. Users cannot send images to vision-capable models (GLM-4.6V, GPT-4o, Claude) through the web interface.

Proposed solution

Add image upload support to Control UI chat:

1. Clipboard paste (Ctrl+V) for images
2. Drag-and-drop image upload
3. File picker button for image selection
4. Properly transmit images to vision-capable models via chat.send API

Alternatives considered

1. Upload images to a URL and share the link (requires external hosting)
2. Use other channels like Telegram, Discord, or WhatsApp (supports images natively)
3. Transfer images to server and use the image tool (inconvenient)

Impact

High - Vision models are becoming mainstream. Users expect to send images through web UI. This feature would significantly improve user experience and make Control UI more competitive with other chat interfaces.

Evidence/examples

Vision-capable models like GPT-4o, Claude 3.5, GLM-4.6V are widely used. All major chat interfaces (ChatGPT, Claude.ai, Gemini) support image uploads. OpenClaw's web UI lacks this basic feature.

Additional information

OpenClaw version: 2026.3.2
Channel: WebChat / Control UI

Tested with GLM-4.6V model configured with image input support. Model works correctly with image URLs via the image tool, but Control UI cannot transmit images directly.