Repository navigation
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
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
A context summary that ended at the model output limit could replace the original request history and enter the summary cache. Later requests then lost requirements that were missing from the partial summary.
Reject the existing normalized truncation reasons before formatting a summary. Single-request summaries use the existing non-cacheable fallback, so the default context shaper retains the original history. Chunked summaries keep their existing failure and transcript recovery behavior. Complete summaries and their cache behavior remain unchanged.
The new regressions reproduce the lost requirement through native OpenAI Chat, Responses and Anthropic SDK transports. They also verify usage accounting, retry and cache behavior, complete stop reasons, and a truncated later chunk. The regression set passed all 47 cases and related context, budget and cancellation tests passed all 270 cases. Repository Ruff, import constraints and dependency checks passed.