[GH-ISSUE #19740] issue: How to pass a file or a file URL when invoking MCP? #18978

New Issue

GiteaMirror · 2026-04-20T01:16:02-05:00

GiteaMirror commented

2026-04-20 01:16:02 -05:00

Originally created by @gongshaojie12 on GitHub (Dec 4, 2025).
Original GitHub issue: https://github.com/open-webui/open-webui/issues/19740

Check Existing Issues

I have searched for any existing and/or related issues.
I have searched for any existing and/or related discussions.
I have also searched in the CLOSED issues AND CLOSED discussions and found no related items (your issue might already be addressed on the development branch!).
I am using the latest version of Open WebUI.

Installation Method

Pip Install

Open WebUI Version

0.6.41

Ollama Version (if applicable)

No response

Operating System

centos7

Browser (if applicable)

No response

Confirmation

I have read and followed all instructions in README.md.
I am using the latest version of both Open WebUI and Ollama.
I have included the browser console logs.
I have included the Docker container logs.
I have provided every relevant configuration, setting, and environment variable used in my setup.
I have clearly listed every relevant configuration, custom setting, environment variable, and command-line option that influences my setup (such as Docker Compose overrides, .env values, browser settings, authentication configurations, etc).
I have documented step-by-step reproduction instructions that are precise, sequential, and leave nothing to interpretation. My steps:
Start with the initial platform/version/OS and dependencies used,
Specify exact install/launch/configure commands,
List URLs visited, user input (incl. example values/emails/passwords if needed),
Describe all options and toggles enabled or changed,
Include any files or environmental changes,
Identify the expected and actual result at each stage,
Ensure any reasonably skilled user can follow and hit the same issue.

Expected Behavior

I created an agent in Dify that requires a file or a file URL as a parameter. After configuring this agent as an MCP tool in Open-WebUI, how can I pass a file or a file URL to the Dify agent when invoking it in Open-WebUI?

I expect to be able to pass a file or a file URL to the Dify agent in Open-WebUI.

Actual Behavior

It is not possible to pass a file or a file URL to the Dify MCP from Open-WebUI.

Steps to Reproduce

None

Logs & Screenshots

None

Additional Information

No response

Originally created by @gongshaojie12 on GitHub (Dec 4, 2025). Original GitHub issue: https://github.com/open-webui/open-webui/issues/19740 ### Check Existing Issues - [x] I have searched for any existing and/or related issues. - [x] I have searched for any existing and/or related discussions. - [x] I have also searched in the CLOSED issues AND CLOSED discussions and found no related items (your issue might already be addressed on the development branch!). - [x] I am using the latest version of Open WebUI. ### Installation Method Pip Install ### Open WebUI Version 0.6.41 ### Ollama Version (if applicable) _No response_ ### Operating System centos7 ### Browser (if applicable) _No response_ ### Confirmation - [x] I have read and followed all instructions in `README.md`. - [x] I am using the latest version of **both** Open WebUI and Ollama. - [x] I have included the browser console logs. - [x] I have included the Docker container logs. - [x] I have **provided every relevant configuration, setting, and environment variable used in my setup.** - [x] I have clearly **listed every relevant configuration, custom setting, environment variable, and command-line option that influences my setup** (such as Docker Compose overrides, .env values, browser settings, authentication configurations, etc). - [x] I have documented **step-by-step reproduction instructions that are precise, sequential, and leave nothing to interpretation**. My steps: - Start with the initial platform/version/OS and dependencies used, - Specify exact install/launch/configure commands, - List URLs visited, user input (incl. example values/emails/passwords if needed), - Describe all options and toggles enabled or changed, - Include any files or environmental changes, - Identify the expected and actual result at each stage, - Ensure any reasonably skilled user can follow and hit the same issue. ### Expected Behavior I created an agent in Dify that requires a file or a file URL as a parameter. After configuring this agent as an MCP tool in Open-WebUI, how can I pass a file or a file URL to the Dify agent when invoking it in Open-WebUI? I expect to be able to pass a file or a file URL to the Dify agent in Open-WebUI. ### Actual Behavior It is not possible to pass a file or a file URL to the Dify MCP from Open-WebUI. ### Steps to Reproduce None ### Logs & Screenshots None ### Additional Information _No response_

GiteaMirror added the bug label 2026-04-20 01:16:02 -05:00

GiteaMirror closed this issue

2026-04-20 01:16:03 -05:00

GiteaMirror commented

2026-04-20 01:16:04 -05:00

@owui-terminator[bot] commented on GitHub (Dec 4, 2025):

🔍 Similar Issues Found

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

#19701 issue: knowledge can not multiple upload file
by willy808 • Dec 03, 2025 • bug
#14726 issue: When the LLM calls mcp tools that return image data, the front UI can not display the image
by tangzhong1 • Jun 06, 2025 • bug
#18981 issue: 0.6.35 error using tools through mcpo
by GlisseManTV • Nov 06, 2025 • bug
#16521 issue: web browser directory access mcpo openapi server
by okamototk • Aug 12, 2025 • bug
#19563 issue:
by naruto7g • Nov 28, 2025 • bug

Show 5 more related issues

#17808 issue: unexpected MCP tool response handling
by dlamoris • Sep 27, 2025 • bug
#15855 issue:
by richtong • Jul 19, 2025 • bug
#14767 issue:
by TheMoye • Jun 07, 2025 • bug
#13451 issue:
by kinnerful • May 03, 2025 • bug
#18041 issue: web search enabled and fetch MCP break rule 2 tools for the same function
by gattytto • Oct 04, 2025 • bug

💡 Tips:

If this is a duplicate, please consider closing this issue and adding any additional details to the existing one
If you found a solution in any of these issues, please share it here to help others

This comment was generated automatically by a bot. Please react with a 👍 if this comment was helpful, or a 👎 if it was not.

@owui-terminator[bot] commented on GitHub (Dec 4, 2025): 🔍 **Similar Issues Found** I found some existing issues that might be related to this one. Please check if any of these are duplicates or contain helpful solutions: 1. [#19701](https://github.com/open-webui/open-webui/issues/19701) **issue: knowledge can not multiple upload file** *by willy808 • Dec 03, 2025 • `bug`* 2. [#14726](https://github.com/open-webui/open-webui/issues/14726) **issue: When the LLM calls mcp tools that return image data, the front UI can not display the image** *by tangzhong1 • Jun 06, 2025 • `bug`* 3. [#18981](https://github.com/open-webui/open-webui/issues/18981) **issue: 0.6.35 error using tools through mcpo** *by GlisseManTV • Nov 06, 2025 • `bug`* 4. [#16521](https://github.com/open-webui/open-webui/issues/16521) **issue: web browser directory access mcpo openapi server** *by okamototk • Aug 12, 2025 • `bug`* 5. [#19563](https://github.com/open-webui/open-webui/issues/19563) **issue:** *by naruto7g • Nov 28, 2025 • `bug`* <details> <summary>Show 5 more related issues</summary> 6. [#17808](https://github.com/open-webui/open-webui/issues/17808) **issue: unexpected MCP tool response handling** *by dlamoris • Sep 27, 2025 • `bug`* 7. [#15855](https://github.com/open-webui/open-webui/issues/15855) **issue:** *by richtong • Jul 19, 2025 • `bug`* 8. [#14767](https://github.com/open-webui/open-webui/issues/14767) **issue:** *by TheMoye • Jun 07, 2025 • `bug`* 9. [#13451](https://github.com/open-webui/open-webui/issues/13451) **issue:** *by kinnerful • May 03, 2025 • `bug`* 10. [#18041](https://github.com/open-webui/open-webui/issues/18041) **issue: web search enabled and fetch MCP break rule 2 tools for the same function** *by gattytto • Oct 04, 2025 • `bug`* </details> --- 💡 **Tips:** - If this is a duplicate, please consider closing this issue and adding any additional details to the existing one - If you found a solution in any of these issues, please share it here to help others *This comment was generated automatically by a bot.* Please react with a 👍 if this comment was helpful, or a 👎 if it was not.

GiteaMirror commented

2026-04-20 01:16:04 -05:00

@gongshaojie12 commented on GitHub (Dec 4, 2025):

The issues you mentioned are different from mine.

@gongshaojie12 commented on GitHub (Dec 4, 2025): The issues you mentioned are different from mine.

GiteaMirror commented

2026-04-20 01:16:05 -05:00

@rgaricano commented on GitHub (Dec 4, 2025):

You can attach it to chat, then it is going to be passed to MCP tools via the __files__ in extra_params.
e.g. for extract it:

async def your_mcp_tool_function(**kwargs):
    # The __files__ parameter contains file metadata
    files = kwargs.get('__files__', [])

    if files:
        for file_item in files:
            # Each file item contains metadata like:
            # - type: "file" or "url"
            # - url: direct URL to the file
            # - name: filename
            # - id: file ID in Open-WebUI

            if file_item.get('type') == 'file':
                file_url = file_item.get('url')
                file_name = file_item.get('name')

                # You can now use the file URL to download or process the file
                # For example, download the content:
                import httpx
                async with httpx.AsyncClient() as client:
                    response = await client.get(file_url)
                    file_content = response.content

                # Process the file content as needed
                # ...

            elif file_item.get('type') == 'url':
                # For URL-type files, the URL points to an external resource
                external_url = file_item.get('url')
                # Fetch and process the external URL content
                # ...

    # Continue with your tool logic
    return {"result": "Files processed successfully"}

Note:
If you cannot see the file list, try extracting them directly from __files__ instead of through kwargs:
async def your_mcp_tool_function(__files__: Optional[dict] = None, **kwargs):
files = __files__ if __files__ is not None else {}

@rgaricano commented on GitHub (Dec 4, 2025): You can attach it to chat, then it is going to be passed to MCP tools via the ` __files__` in `extra_params`. e.g. for extract it: ``` async def your_mcp_tool_function(**kwargs): # The __files__ parameter contains file metadata files = kwargs.get('__files__', []) if files: for file_item in files: # Each file item contains metadata like: # - type: "file" or "url" # - url: direct URL to the file # - name: filename # - id: file ID in Open-WebUI if file_item.get('type') == 'file': file_url = file_item.get('url') file_name = file_item.get('name') # You can now use the file URL to download or process the file # For example, download the content: import httpx async with httpx.AsyncClient() as client: response = await client.get(file_url) file_content = response.content # Process the file content as needed # ... elif file_item.get('type') == 'url': # For URL-type files, the URL points to an external resource external_url = file_item.get('url') # Fetch and process the external URL content # ... # Continue with your tool logic return {"result": "Files processed successfully"} ``` Note: If you cannot see the file list, try extracting them directly from `__files__` instead of through `kwargs`: `async def your_mcp_tool_function(__files__: Optional[dict] = None, **kwargs):` `files = __files__ if __files__ is not None else {}`

GiteaMirror commented

2026-04-20 01:16:05 -05:00

@gongshaojie12 commented on GitHub (Dec 4, 2025):

Where should this code be placed? My MCP is configured like this, as shown below:

@gongshaojie12 commented on GitHub (Dec 4, 2025): Where should this code be placed? My MCP is configured like this, as shown below: <img width="1182" height="873" alt="Image" src="https://github.com/user-attachments/assets/cbd4ea6c-21e7-4429-b4ca-a018e85330a3" />