fix(core,cli): label screenshot-triggered compaction accurately in the auto-compact notice - #4623
Merged
Merged
Conversation
…e auto-compact notice
The auto-compaction notice hardcoded "approached the input token limit"
even when the screenshot-overflow trigger fired. In computer-use sessions
that's misleading: compaction can fire on accumulated tool screenshots
while token usage is far below the window limit (observed: the notice
claimed "approached the input token limit" at ~116K/1M tokens when it was
actually the image-count trigger).
Add ChatCompressionInfo.triggerReason ('token_limit' | 'image_overflow' |
'manual'); compress() sets it to 'image_overflow' when the screenshot
trigger is what let it through the cheap gate. Both the TUI
(useGeminiStream) and ACP (Session) notices now show an accurate clause.
Contributor
📋 Review SummaryThis PR addresses a user-facing accuracy issue where the auto-compaction notice incorrectly reported "approached the input token limit" when compaction was actually triggered by screenshot/image overflow. The implementation threads a 🔍 General Feedback
🎯 Specific Feedback🟢 Medium
const reasonClause =
compressed.triggerReason === 'image_overflow'
? `accumulated enough tool screenshots...`
: compressed.triggerReason === 'manual'
? `manually triggered compaction for...`
: `approached the input token limit for...`;🔵 Low
export type CompactionTriggerReason =
| 'token_limit' // Auto-triggered when effective tokens exceed threshold
| 'image_overflow' // Auto-triggered by screenshot count despite being below token threshold
| 'manual'; // User-initiated via explicit compact command
function buildCompressionNotice(
triggerReason: CompactionTriggerReason | undefined,
model: string,
originalCount: number | undefined,
newCount: number | undefined,
): string {
const reasonClause = triggerReason === 'image_overflow'
? `accumulated enough tool screenshots to trigger compaction for ${model}`
: `approached the input token limit for ${model}`;
return `IMPORTANT: This conversation ${reasonClause}. A compressed context...`;
}
✅ Highlights
|
pomelo-nwu
approved these changes
May 29, 2026
Contributor
Code Coverage Summary
CLI Package - Full Text ReportCore Package - Full Text ReportFor detailed HTML reports, please see the 'coverage-reports-22.x-ubuntu-latest' artifact from the main CI run. |
wenshao
approved these changes
May 29, 2026
wenshao
left a comment
wenshao
left a comment
Collaborator
There was a problem hiding this comment.
No issues found. LGTM! ✅ — qwen3.7-max via Qwen Code /review
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
4 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.
Summary
The auto-compaction notice always read "This conversation approached the input token limit for {model}", even when compaction was actually fired by the screenshot/image-overflow trigger added in #4599. With the trigger threshold set well below the token-auto threshold (e.g. 20 images on a 1M-window model), the notice was misleading — the conversation had nowhere near the token limit; it had simply accumulated enough tool screenshots.
This PR threads a
triggerReasonthrough the compaction result so the notice reflects what actually happened.Changes
CompactionTriggerReason = 'token_limit' | 'image_overflow' | 'manual'and an optionaltriggerReasonfield onChatCompressionInfo.triggerReasonincompress(): defaults tomanual/token_limitby trigger, upgraded toimage_overflowin the gate when the screenshot trigger is what let a below-threshold compaction through. Surfaced on theCOMPRESSEDresult.triggerReason. Screenshot-triggered compaction reads "accumulated enough tool screenshots to trigger compaction for {model}"; the token-limit wording is unchanged.The notice string is not i18n'd today, so this stays English-only to match the existing behavior.
Test plan
chatCompressionService.test.ts— asserttriggerReason === 'image_overflow'on the two screenshot-trigger COMPRESSED cases and'token_limit'on the token-trigger case.Session.test.ts— new case asserting the ACP notice reads "accumulated enough tool screenshots…" whentriggerReason === 'image_overflow'.tsc --noEmitclean.