fix: strip CacheControl from messages during compaction#2236
Merged
dgageot merged 1 commit intoMar 25, 2026
Merged
Conversation
When compaction runs, the original session's system messages carry CacheControl=true markers. These were preserved through BuildPrompt into the summarization session, which then added its own markers plus applyMessageCacheControl's 2 markers, exceeding Anthropic's limit of 4 cache control breakpoints. Assisted-By: docker-agent
trungutt
approved these changes
Mar 25, 2026
Sayt-0
pushed a commit
that referenced
this pull request
Jul 17, 2026
… the limit The 4-breakpoint budget was an implicit cross-layer contract: the session marks up to two system messages and the provider spends the rest, but nothing enforced the global invariant \u2014 a third upstream mark (e.g. from a hook-rewritten message list) would produce a request Anthropic rejects outright (the failure mode of #2236). Own the whole budget in cache_control.go: document the ledger and cap honored system marks at maxSystemCacheBreakpoints, dropping the excess instead of forwarding it. Behavior is unchanged for compliant input.
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
2 participants
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
When compaction runs,
doCompactcallssess.GetMessages(a)which returns messages withCacheControl=trueon the last invariant and last context system messages. These markers were preserved throughcompaction.BuildPromptinto the summarization session, which then added its own system message marker plus 2 more fromapplyMessageCacheControl, totaling 5 cache control breakpoints — exceeding Anthropic's limit of 4.Fix: Strip
CacheControlfrom cloned messages inBuildPrompt, just likeCostis already zeroed.