mirror of
https://github.com/open-webui/open-webui.git
synced 2026-07-17 08:21:12 -05:00
[GH-ISSUE #24444] feat: Tool Callbacks #107300
No Assignees
Notifications
Due Date
No due date set.
Dependencies
No dependencies set.
Reference: github-starred/open-webui#107300
Reference in New Issue
Block a user
Delete Branch "%!s()"
Deleting a branch is permanent. Although the deleted branch may continue to exist for a short time before it actually gets removed, it CANNOT be undone in most cases. Continue?
Originally created by @Notbici on GitHub (May 8, 2026).
Original GitHub issue: https://github.com/open-webui/open-webui/issues/24444
Check Existing Issues
Verify Feature Scope
Problem Description
Sometimes I want my LLM to perform a task that requires some action to happen at a later date. Current tool calls require input and an output immediately.
Desired Solution you'd like
I'd love for tool calls to return a callback ID, then via webhook or other medium/api I can have my tool respond to that callback ID with the output and have that result get sent to the chat live for the LLM to process.
This means the chat can continue while we're waiting on the output to cook, and once its done the LLM can process and respond.
Alternatives Considered
Not sure if one exists easily?
Additional Context
I wanted my LLM to execute an update sequence to a server I'm working on via openwebui tools. Instead of the tool locking up the chat I wanted it to run in the background then callback when its finished.
Honestly maybe it doesn't even need the 'callback' api endpoint, just the fact that the tool executes in the background in parallel?
@owui-terminator[bot] commented on GitHub (May 8, 2026):
🔍 Related Issues Found
I found some existing issues that might be related. Please check if any of these are duplicates or contain helpful solutions:
🟣 #22565 issue: Message queue should not block new messages while background tasks are still running
This is the closest workflow match: it asks for background tasks to run without blocking the main chat flow. The new issue similarly wants tool work to continue asynchronously so the conversation can proceed while results are still pending.
by ShirasawaSama ·
bug🟣 #11288 issue: functions doesn't work asynchronously
This issue reports that functions/filters do not work asynchronously and fail when the tab is closed, which is directly related to the broader need for background execution independent of the active chat request.
by adan89lion ·
bug🟣 #12735 issue: Logging tool results into chat
This request asks for tool results to be logged back into chat so the LLM can use them in later turns. That overlaps with the new feature’s goal of delivering tool output back to the conversation after a delay.
by bmen25124 ·
bug🟣 #9755 Support for multiple tool execution
Although framed as multiple tool execution, it is about tool orchestration and getting more than one tool result into the LLM context. It is relevant because the callback feature is another way to support more flexible tool completion behavior.
by rodrigopv
🟣 #19783 feat: Models Called from API automatically invoke Open-webUI tools, not respond with tool request code
This issue is about tool-calling flow and automatic handling of tool requests versus returning intermediate tool-call messages. It’s related because the new feature also changes how tool calls are completed and fed back into the chat/API flow.
by dwallner0554966
💡 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.
@Classic298 commented on GitHub (May 8, 2026):
you can build a tool to create the task a tool to wait to have the LLM literally wait in the chat and another tool where the AI specifies the ID to the check the status of the task that runs elsewhere. Implementable via tools