bug: agent can not see images. display_file only gives agent plain text path to file

[pixel4696]

it seems that agent can not see images. display_file only gives agent plain text path to file.

i am using openai endpoint from open-webui, which gathers my self hosted VLMs that do support image recognition (works in open-webui)

[owui-terminator]

⚠️ Missing Issue Title Prefix

@pixel4696, your issue title is missing a categorising prefix (e.g. bug:, feat:, docs:).

Please update the title to lead with one of:

• bug: bug report or error
• feat: feature request or enhancement
• docs: documentation issue
• question: usage question
• help: support request

Example: bug: Login fails when the password contains special characters

[owui-terminator]

🔍 Related Issues Found

I found some existing issues that might be related. Please check if any of these are duplicates or contain helpful solutions:

1. 🟢 #246 bug: image previews never render on Windows — frontend sends backslash paths to /api/workspace/files/serve
   This issue is directly about display_file/workspace file serving for images. It reports that the frontend passes an invalid raw path to /api/workspace/files/serve, causing image display to fail even though the backend can serve the same file correctly with a normalized URL.
   by gucazcode

2. 🟣 #248 Issue bug: File preview serves stale content for unchanged file paths #248
   This is in the same file preview/display path and concerns preview behavior around workspace files. It may be relevant because the new issue also centers on file display through display_file, although the symptom here is stale content rather than missing vision access.
   by unknown

---

💡 If your issue is a duplicate, please close it and add any additional details to the existing issue instead.

This comment was generated automatically. React with 👍 if helpful, 👎 if not.