Teams channel bridge drops Ctrl+V pasted images — Azure AI Foundry Hosted Agent (Responses protocol) only receives text

Question

Teams channel bridge drops Ctrl+V pasted images — Azure AI Foundry Hosted Agent (Responses protocol) only receives text

Mohamed, Rihan 60

Issue Summary

When interacting with an Azure AI Foundry Hosted Agent published to Microsoft Teams, images uploaded using the Attach (📎) button are processed correctly. However, images pasted directly into the Teams chat using Ctrl+V are not forwarded to the hosted agent.

As a result, the agent receives only the text prompt and cannot analyze or summarize the pasted image.

Environment

Azure AI Foundry Hosted Agent (Python)
Protocol: Responses
Published to Microsoft Teams through Azure AI Foundry
Vision-capable model (GPT-4o / GPT-5)

Working Scenario (Image Upload)

When an image is uploaded using the Teams Attach (📎) button, the request received by the hosted agent contains both the text prompt and the image.

Example log:

The agent successfully processes the image and returns the expected response.

Failing Scenario (Ctrl+V Paste)

When the same image is pasted into the Teams chat using Ctrl+V, the hosted agent receives only the text prompt.

Example log:

No image content is included in the request. Specifically, there is no:

input_file
input_image
image_url
Base64 image data

Because the image is missing from the request, the agent responds that no image was received.

Investigation Findings

We investigated the issue using Application Insights and custom logging.

Image uploaded via Attach button

The hosted agent receives:

This confirms that uploaded images are correctly forwarded to the Responses API.

Image pasted using Ctrl+V

The hosted agent receives only:

The image is not present anywhere in the request received by the hosted agent.

This indicates that the image is lost before the request reaches the agent.

Expected Behavior

Images pasted into Microsoft Teams using Ctrl+V should be forwarded to Azure AI Foundry Hosted Agents in the same way as images uploaded using the Attach (📎) button.

The Responses API request should include the pasted image as an input_file (or equivalent multimodal input), allowing the vision model to process it.

Actual Behavior

Attach (📎) upload: ✅ Image reaches the hosted agent and is processed successfully.
Ctrl+V pasted image: ❌ Only the text prompt reaches the hosted agent. The image is not included in the request.

Impact

Users naturally paste screenshots using Ctrl+V in Microsoft Teams. Since these images are not forwarded to Azure AI Foundry Hosted Agents, vision capabilities cannot be used unless users explicitly upload the image via the Attach button.

Request

Please investigate the Teams-to-Foundry integration for Hosted Agents using the Responses protocol and verify whether clipboard-pasted images (image/png, image/jpeg, etc.) are expected to be forwarded to the hosted agent.

If this scenario is supported, please identify why pasted images are not being included in the Responses API request. If it is not currently supported, please confirm whether this is a known limitation or whether support for pasted images is planned.

Issue Summary

When interacting with an Azure AI Foundry Hosted Agent published to Microsoft Teams, images uploaded using the Attach (📎) button are processed correctly. However, images pasted directly into the Teams chat using Ctrl+V are not forwarded to the hosted agent.

As a result, the agent receives only the text prompt and cannot analyze or summarize the pasted image.

Environment

Azure AI Foundry Hosted Agent (Python)
Protocol: Responses
Published to Microsoft Teams through Azure AI Foundry
Vision-capable model (GPT-4o / GPT-5)

Working Scenario (Image Upload)

When an image is uploaded using the Teams Attach (📎) button, the request received by the hosted agent contains both the text prompt and the image.

Example log:

The agent successfully processes the image and returns the expected response.

Failing Scenario (Ctrl+V Paste)

When the same image is pasted into the Teams chat using Ctrl+V, the hosted agent receives only the text prompt.

Example log:

No image content is included in the request. Specifically, there is no:

input_file
input_image
image_url
Base64 image data

Because the image is missing from the request, the agent responds that no image was received.

Investigation Findings

We investigated the issue using Application Insights and custom logging.

Image uploaded via Attach button

The hosted agent receives:

This confirms that uploaded images are correctly forwarded to the Responses API.

Image pasted using Ctrl+V

The hosted agent receives only:

The image is not present anywhere in the request received by the hosted agent.

This indicates that the image is lost before the request reaches the agent.

Expected Behavior

Images pasted into Microsoft Teams using Ctrl+V should be forwarded to Azure AI Foundry Hosted Agents in the same way as images uploaded using the Attach (📎) button.

The Responses API request should include the pasted image as an input_file (or equivalent multimodal input), allowing the vision model to process it.

Actual Behavior

Attach (📎) upload: ✅ Image reaches the hosted agent and is processed successfully.
Ctrl+V pasted image: ❌ Only the text prompt reaches the hosted agent. The image is not included in the request.

Impact

Users naturally paste screenshots using Ctrl+V in Microsoft Teams. Since these images are not forwarded to Azure AI Foundry Hosted Agents, vision capabilities cannot be used unless users explicitly upload the image via the Attach button.

Request

Please investigate the Teams-to-Foundry integration for Hosted Agents using the Responses protocol and verify whether clipboard-pasted images (image/png, image/jpeg, etc.) are expected to be forwarded to the hosted agent.

If this scenario is supported, please identify why pasted images are not being included in the Responses API request. If it is not currently supported, please confirm whether this is a known limitation or whether support for pasted images is planned.

Deleted

This comment has been deleted due to a violation of our Code of Conduct. The comment was manually reported or identified through automated detection before action was taken. Please refer to our Code of Conduct for more information.
Deleted

This comment has been deleted due to a violation of our Code of Conduct. The comment was manually reported or identified through automated detection before action was taken. Please refer to our Code of Conduct for more information.
Anshika Varshney 15,625 Reputation points Microsoft External Staff Moderator

2026-07-15T21:54:47.1566667+00:00

Hello @Mohamed, Rihan

We haven’t heard from you on the last response and was just checking back to see if you have a resolution yet. In case if you have any resolution please do share that same with the community as it can be helpful to others. Otherwise, will respond with more details and we will try to help.

Thankyou!
Anshika Varshney 15,625 Reputation points Microsoft External Staff Moderator

2026-07-16T20:19:34.41+00:00

Hello @Mohamed, Rihan

If these steps resolve your issue, please mark this answer as Accepted so others can easily find the solution.

If you still face problems, share additional details and we’ll assist further.

Thankyou!

Answer accepted by question author

Anshika Varshney 15,625 Microsoft External Staff Moderator

Hello @Mohamed, Rihan

Thank you for the detailed investigation and for sharing the request logs.

Based on the information provided, the behavior appears to occur before the request reaches the Azure AI Foundry Hosted Agent. Your logs show that images uploaded through the Teams Attach button are successfully forwarded to the agent, while images pasted using Ctrl+V are not included in the request payload at all. The hosted agent receives only the text prompt, without any input_file, input_image, image_url, or image data.

Since the uploaded-image scenario works correctly, this suggests the agent, model deployment, and Responses protocol are functioning as expected. The difference appears to be in how Microsoft Teams handles clipboard-pasted images versus attached files before forwarding content to the Foundry agent integration.

As a troubleshooting step, please verify whether the same behavior occurs:

In both Teams desktop and Teams web clients.
Across multiple users and Teams tenants.
With different image formats and screenshot sources.

At present, the information provided indicates that only images uploaded through the Attach workflow are consistently reaching the hosted agent, whereas clipboard-pasted images are not being forwarded.

Given that the image is absent from the request received by the agent, further investigation from the product team would be required to determine whether this is:

A known limitation of the Teams-to-Foundry integration.
A regression in the current implementation.
An unsupported scenario for clipboard-pasted images.

In the meantime, the recommended workaround is to upload images using the Attach option, which you have already confirmed successfully forwards image content to the hosted agent.

Could you also confirm whether this behavior is reproducible in both Teams desktop and web clients, and whether it affects all hosted agents or only a specific agent deployment?

Thankyou!

0 comments

1 additional answer

Your answer

Deleted

This comment has been deleted due to a violation of our Code of Conduct. The comment was manually reported or identified through automated detection before action was taken. Please refer to our Code of Conduct for more information.
Anshika Varshney 15,625 Reputation points Microsoft External Staff Moderator

2026-07-15T21:54:47.1566667+00:00

Hello @Mohamed, Rihan

We haven’t heard from you on the last response and was just checking back to see if you have a resolution yet. In case if you have any resolution please do share that same with the community as it can be helpful to others. Otherwise, will respond with more details and we will try to help.

Thankyou!
Anshika Varshney 15,625 Reputation points Microsoft External Staff Moderator

2026-07-16T20:19:34.41+00:00

Hello @Mohamed, Rihan

If these steps resolve your issue, please mark this answer as Accepted so others can easily find the solution.

If you still face problems, share additional details and we’ll assist further.

Thankyou!

Answer 1

Christos Panagiotidis 3,551

Hi! Your logging demonstrates that pasted images and attached files are represented differently by Teams, and the current channel bridge is forwarding only the attachment form. There is no agent-side fix if the image never appears in the incoming request. Continue using the Attach button as the workaround and report this as a Teams/Foundry channel connector limitation, including a conversation ID from one pasted-image test and one successful attachment test.

0 comments