You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
When I use the adk web command to launch the agent through the web UI and ask the agent to help me write Python code. I use matplotlib to draw and display the image, the image can be successfully generated but the web UI cannot display it. May I ask if this is not supported by the web UI or if I need to enable additional parameters?
You are asking about generated images from code executors. Could you check the artifact tab and see if your image is there? For now the <img src="link" /> is probably the easiest way to get around with a simple prompt change
it would be nice to have this feature in place. Even if you upload an image, and then you ask the agent to show it it would say something like:
I apologize, but I am unable to display the image again. I can confirm that I have access to the image and can describe its contents, but I cannot visually render it within this text-based environment.
Can you share the code for generating images? @Tyrooon
I am not sure how to pass the image to the agent, to be able to show it on UI.
If you want to upload an image, just click on the attachment button in the input box
I have another problem. When I use the attachment button to upload an image, ADK will pass the image to LLM directly. How can I just pass the attachment to my dictionary?
Activity
ShawnxWX commentedon May 1, 2025
one use case using markdown image tag with link to the image
another way is in the prompt ask LLM to render imageUrl as html tag
<img src="link" />wyf7107 commentedon May 2, 2025
You are asking about generated images from code executors. Could you check the artifact tab and see if your image is there? For now the
<img src="link" />is probably the easiest way to get around with a simple prompt changeAntriksh-Narang commentedon May 8, 2025
Can you share the code for generating images? @Tyrooon
I am not sure how to pass the image to the agent, to be able to show it on UI.
wyf7107 commentedon May 9, 2025
If you want to upload an image, just click on the attachment button in the input box
jonigl commentedon May 14, 2025
it would be nice to have this feature in place. Even if you upload an image, and then you ask the agent to show it it would say something like:
Tyrooon commentedon May 16, 2025
I added
<img src="link" />in the description of my agent and solve the problem.Tyrooon commentedon May 16, 2025
I have another problem. When I use the attachment button to upload an image, ADK will pass the image to LLM directly. How can I just pass the attachment to my dictionary?
rajashish commentedon May 20, 2025
@Tyrooon can you please share the code which is able to display image in the adk web. I am trying with the img src tag but still its not working
boyangsvl commentedon May 21, 2025
@Tyrooon you can add a callback to read teh uploaded image and put it in wherever you want it to be.
I think the generated artifact will display in adk web automatically. Displaying of external images may be restricted due to security vulnerability.
I will move this issue to discussions if you want to continue the discussion.