🤖 fix: prevent queued message amnesia by flushing partial before tool-call-end #807
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
…-call-end When a queued message is sent after tool-call-end, the model needs to see the complete assistant response including tool results. Previously, there was a race condition where tool-call-end was emitted before the partial file was written, causing commitToHistory to read stale data. The fix ensures partial.json is flushed to disk BEFORE emitting the tool-call-end event, so listeners (like sendQueuedMessages) see the complete tool result when they read via commitToHistory. Changes: - Make completeToolCall async - Await flushPartialWrite before emitting tool-call-end - Update callers to await the async method _Generated with `mux`_
ethanndickson
force-pushed
the
fix-queued-message-amnesia
branch
from
fb8c7ce to
058d93b
Compare
Member
Author
|
We still buffer partial writes for |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Problem
When a queued message is sent after
tool-call-end, the model doesn't see the complete assistant response including tool results. This causes "amnesia" where the model starts tasks from scratch.Root Cause
Race condition in partial write timing:
The
schedulePartialWritewas called withvoid(fire-and-forget). WhensendQueuedMessagestriggerscommitToHistory, the partial file may not contain the latest tool results.Solution
Await the partial flush before emitting
tool-call-end:Changes
completeToolCallasyncflushPartialWritebefore emittingtool-call-endGenerated with
mux