fix(core): preserve reasoning metadata on errored assistant turns#38763
Open
loulanyue wants to merge 1 commit into
Open
fix(core): preserve reasoning metadata on errored assistant turns#38763loulanyue wants to merge 1 commit into
loulanyue wants to merge 1 commit into
Conversation
Fixes anomalyco#38620. Anthropic rejects a replayed `tool_use` if its preceding `thinking` block is dropped. By relaxing the `message.error === undefined` guard for reasoning blocks, we ensure the signature remains and the block is not filtered out as empty, keeping the message valid for the provider.
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
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Issue for this PR
Closes #38620
Type of change
What does this PR do?
When a session turn is interrupted (network error, timeout, etc.) and contains both a
thinkingblock and atool_useblock, the existing replay logic into-llm-message.tsstrippedproviderMetadatafrom thereasoningblock becausemessage.errorwas defined. This left thetool_useblock without a valid precedingthinkingblock, causing Anthropic to return a 400 error on replay.The fix is minimal: always pass
item.providerMetadataforreasoningblocks, regardless of whether the message has an error. ThereuseProviderMetadataflag still guards tool-related metadata on errored turns; only reasoning metadata is now unconditionally preserved.How did you verify your code works?
Updated the existing test
drops provider-native continuation metadata from failed assistant turns(now renamed to reflect the new behavior) to assert that reasoning blocks retain their providerMetadata while tool-related continuation metadata is still dropped. All session tests pass:Screenshots / recordings
No UI changes.
Checklist