fix(vscode-lm): reliable auto context condensing#710
Merged
Conversation
Port of simurg79/Roo-Code#11 into Zoo-Code.
simurg79
requested review from
JamesRobert20,
edelauna,
hannesrudolph,
navedmerchant and
taltas
as code owners
Contributor
|
Note Reviews pausedIt looks like this branch is under active development. To avoid overwhelming you with review comments due to an influx of new commits, CodeRabbit has automatically paused this review. You can configure this behavior by changing the Use the following commands to manage reviews:
Use the checkboxes below for quick actions:
📝 WalkthroughWalkthroughUpdates the VS Code LM model catalog, adds a ChangesVS Code LM condense gate fix
Estimated code review effort🎯 4 (Complex) | ⏱️ ~60 minutes Possibly related PRs
Suggested labels
Suggested reviewers
Poem
🚥 Pre-merge checks | ✅ 4 | ❌ 1❌ Failed checks (1 warning)
✅ Passed checks (4 passed)
✨ Finishing Touches🧪 Generate unit tests (beta)
Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out. Comment |
![coderabbitai[bot]](https://avatars.githubusercontent.com/in/347564?s=60&v=4){kind=small}
Contributor
There was a problem hiding this comment.
Actionable comments posted: 1
🤖 Prompt for all review comments with AI agents
Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.
Inline comments:
In @.changeset/vscode-lm-condense-fix.md:
- Around line 1-5: The changeset file `.changeset/vscode-lm-condense-fix.md` was
created but should not be included in this PR since changesets are managed by
maintainers outside the normal development workflow. Delete the entire
`.changeset/vscode-lm-condense-fix.md` file and allow maintainers to create the
proper changeset entry separately.
🪄 Autofix (Beta)
Fix all unresolved CodeRabbit comments on this PR:
- Push a commit to this branch (recommended)
- Create a new PR with the fixes
ℹ️ Review info
⚙️ Run configuration
Configuration used: defaults
Review profile: CHILL
Plan: Pro Plus
Run ID: 011fdda6-5a92-4a6e-a80c-e1b4d20c91ab
📒 Files selected for processing (13)
.changeset/vscode-lm-condense-fix.mdpackages/types/src/__tests__/vscode-llm.spec.tspackages/types/src/providers/vscode-llm.tssrc/api/index.tssrc/api/providers/__tests__/vscode-lm.spec.tssrc/api/providers/vscode-lm.tssrc/core/context-management/__tests__/context-management.spec.tssrc/core/context-management/index.tssrc/core/task/Task.tswebview-ui/src/components/chat/TaskHeader.tsxwebview-ui/src/components/chat/__tests__/TaskHeader.spec.tsxwebview-ui/src/components/ui/hooks/__tests__/useSelectedModel.spec.tswebview-ui/src/components/ui/hooks/useSelectedModel.ts
Codecov Report❌ Patch coverage is
📢 Thoughts on this report? Let us know! |
Add targeted tests for the previously-uncovered ported branches: the availableInputTokens<=0 fallback to 100% in willManageContext/manageContext, getCondenseContextWindow() guard fallbacks, and the vscode-lm UI family-miss window resolution. Raises patch coverage to satisfy the codecov/patch 80% gate.
Contributor
There was a problem hiding this comment.
Actionable comments posted: 1
🤖 Prompt for all review comments with AI agents
Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.
Inline comments:
In `@src/api/providers/__tests__/vscode-lm.spec.ts`:
- Around line 504-521: The test is mutating the static model row for the
selector family, but the mocked client currently uses a different family so
VsCodeLmHandler never consults that row. Update the test setup in the
VsCodeLmHandler/getCondenseContextWindow case so the mockLanguageModelChat
family matches "claude-opus-4.8", or avoid assigning client so the selector
family path is used; this ensures the zeroed maxInputTokens row is actually
exercised and the live-window fallback assertion is valid.
🪄 Autofix (Beta)
Fix all unresolved CodeRabbit comments on this PR:
- Push a commit to this branch (recommended)
- Create a new PR with the fixes
ℹ️ Review info
⚙️ Run configuration
Configuration used: defaults
Review profile: CHILL
Plan: Pro Plus
Run ID: 7cff98c0-cb87-4d62-9e6d-29e09bf5e9b0
📒 Files selected for processing (2)
src/api/providers/__tests__/vscode-lm.spec.tssrc/core/context-management/__tests__/context-management.spec.ts
…w guard test - Remove .changeset/vscode-lm-condense-fix.md (changesets are maintainer-managed per AGENTS.md; CodeRabbit flagged). - Fix getCondenseContextWindow() non-positive-guard test so the selector family (claude-opus-4.8) drives the lookup and the zeroed static row actually exercises the maxInputTokens > 0 guard before falling back.
edelauna
requested changes
Jun 25, 2026
edelauna
left a comment
edelauna
left a comment
Contributor
There was a problem hiding this comment.
Thanks for addressing this, had some comments around the implementation.
github-actions
Bot
added
the
awaiting-author
…lm; address review Address review feedback from edelauna on Zoo-Code-Org#710: - Scope the available-input-space condense percent denominator to vscode-lm only (via the getCondenseContextWindow seam); all other providers keep dividing by the full context window. The maxTokens:-1 reserve guard remains global. - Correct the misleading useSelectedModel comment: the gate's primary window is getCondenseContextWindow() (static maxInputTokens), not getModel().info.contextWindow. - Strengthen the listed-family test with a claude-opus-4.8 case (contextWindow != maxInputTokens) to catch a field swap.
Simplify comments added in PR Zoo-Code-Org#710 to be brief and rationale-focused; no logic, assertions, or test values changed.
edelauna
requested changes
Jun 29, 2026
edelauna
left a comment
edelauna
left a comment
Contributor
There was a problem hiding this comment.
Thanks for adding this, just had some minor nits.
added 3 commits
June 29, 2026 08:55
Address review: make the negative-reserve percent test discriminate the maxTokens>0 guard (-200000), and add a manageContext truncation-path test so both copies of the -1 reserve guard are covered. Test-only; no product code changed.
Mock utils/storage and OutputInterceptor in Task.dispose.test.ts so dispose()'s unawaited cleanup chain cannot emit a late console.error during Vitest worker teardown (EnvironmentTeardownError: Closing rpc while onUserConsoleLog was pending). Test-only; no product code changed.
…rcent helper getCondenseContextWindow() resolves the default vscode-lm row for an unknown/absent family (catalog drift) instead of the inflated live window; only a non-positive static maxInputTokens still falls back to it. Extract the duplicated contextPercent math shared by willManageContext and manageContext into computeContextPercent so the two stay in lockstep. Addresses PR review feedback.
github-actions
Bot
added
awaiting-review
edelauna
approved these changes
Jul 1, 2026
edelauna
left a comment
edelauna
left a comment
Contributor
There was a problem hiding this comment.
Thanks for addressing the feedback and your contribution!
hacker-b2k
pushed a commit
to hacker-b2k/Zoo-Code
that referenced
this pull request
Jul 2, 2026
…imported from upstream 9a2e8d8]
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.
Related GitHub Issue
Closes #714
Description
Fixes unreliable automatic context condensing on the VS Code LM (
vscode-lm) provider. The provider reportsmaxTokens: -1(unlimited) and an inflated live context window, so the auto-condense gate computed usage against the wrong denominator and effectively never fired even when the UI context gauge showed the window as full.maxTokens: -1(unlimited) as the default output reserve in bothwillManageContextandmanageContextinstead of letting a negative value distort the window math.contextWindow - reservedForOutput), matching the UI gauge, with a safe fallback to the full window when the reserve is unknown/unlimited.getCondenseContextWindow()ApiHandlerseam; only the vscode-lm provider overrides it to drive the gate from the curated static-tablemaxInputTokens. Every other provider falls back tomodelInfo.contextWindow(behavior unchanged).claude-sonnet-4.5).maxTokensas zero reserve and resolves an unlisted family to the default-model window.Test Procedure
packages/types→vscode-llm.spec.ts: model catalog invariants.src→context-management.spec.ts: reserve guard + available-input denominator, including theavailableInputTokens <= 0 → 100%fallback and end-to-endmanageContextsummarization.src→vscode-lm.spec.ts:getCondenseContextWindow()resolution (static family, live fallback, non-positive guard).webview-ui→TaskHeader.spec.tsx,useSelectedModel.spec.ts: UI reserve/window guards.codecov/patchat 88.88% (≥ 80% target).Pre-Submission Checklist
Documentation Updates
Additional Notes
Port of simurg79/Roo-Code#11 into Zoo-Code. Applied by context (paths map 1:1); the upstream version bump was intentionally omitted.
Summary by CodeRabbit
Summary
New Features
Bug Fixes
Tests