openclaw - 💡(How to fix) Fix Bug: Control UI image attachments not passed to Agent [1 participants]

When sending image attachments through the OpenClaw Control UI (webchat channel), the images are successfully uploaded and saved to media/inbound/ directory, but the Agent does not receive the image data. The image tool fails with image required error.

Fix Plan

To resolve the issue, we need to modify the chat.send RPC handler to include MediaPaths and MediaTypes in the ctx object.

Code Changes

Here are the steps to update the code:

• Locate the chat.send RPC handler function.
• After the image is persisted, add the following lines to include MediaPaths and MediaTypes in the ctx object:

ctx.MediaPaths = [image_path]  # assuming image_path is the path where the image is saved
ctx.MediaTypes = ['image/jpeg']  # assuming the image type is JPEG

• Update the dispatchInboundMessage function call to pass the updated ctx object.

Example code snippet:

def chat_send_rpc_handler(ctx, message):
    # ... (image upload and persistence code)
    image_path = 'media/inbound/image.jpg'
    ctx.MediaPaths = [image_path]
    ctx.MediaTypes = ['image/jpeg']
    dispatchInboundMessage(ctx, message)

Verification

To verify the fix, send an image attachment through the OpenClaw Control UI and check if the Agent receives the image data correctly.

Extra Tips

Make sure to handle cases where multiple images are uploaded and update the MediaPaths and MediaTypes lists accordingly. Additionally, consider adding error handling for cases where image persistence fails.