🤖 fix: don't persist compaction model as workspace AI settings #1259
+81
−34
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Outdated suggestions cannot be applied.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 compaction is triggered with a specific model (e.g.,
/compact -m openai:gpt-4.1-mini), the backend was incorrectly persisting that model to the workspace's AI settings, overwriting the user's preferred model.After compaction completed, the workspace would be set to use the compaction model instead of the model the user had selected.
Root Cause
Regression introduced in 1ae0377 (#1203) which added a "safety net" to persist AI settings from
sendMessage/resumeStreamfor cross-device consistency. This safety net didn't account for compaction requests which intentionally use a different (often cheaper/faster) model.cc @tomaskoubek (regressor)
Fix
Skip persisting AI settings when:
sendMessageis called withmuxMetadata.type === "compaction-request"resumeStreamis called withoptions.mode === "compact"Testing
updateAISettingstest still passesGenerated with
mux• Model:anthropic:claude-opus-4-5• Thinking:high