[Anish Mehta (anishmehta24)]

Motivation & Context

_convert_agui_content_to_framework only treats binary, image, audio, video and document parts as media. An input_image part fell through to the text fallback, so the model received the Python repr of the dict as text instead of the image:

_convert_agui_content_to_framework([{"type": "input_image", "source": {"type": "url", "url": "https://example.com/img.png"}}])
# [Content(type="text", text="{'type': 'input_image', 'source': {...}}")]

_normalize_snapshot_content in the same file already maps input_image to a media part, so the two paths disagreed.

Description & Review Guide

• What are the major changes? Treat an input_image part as image before the media check, so it goes through _parse_multimodal_media_part like other images.
• What is the impact of these changes? Image parts sent as input_image reach the model as image content.
• What do you want reviewers to focus on? It reuses the existing image handling, so URL, base64 and binary-id sources all work the same as for image.

Related Issue

No existing issue.

Contribution Checklist

• The code builds clean without any errors or warnings
• All unit tests pass, and I have added new tests where possible

Test: test_convert_content_input_image_is_media_not_text in tests/ag_ui/test_message_adapters.py; it fails on main. The ag-ui tests pass and the repo's pre-commit checks are clean.

[Copilot]

Copilot was unable to review this pull request because the user who requested the review has reached their quota limit.