fix(llmloop): finalize exhausted reviews - #765
Open
4-1-1 wants to merge 1 commit into
Open
Conversation
Give models terminal-only rounds after context exhaustion or a text-only response, while preserving explicit task_done completion.
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Description
When a per-file review reaches its context-tool round budget, the loop currently stops immediately unless the model already called task_done. A text-only response also retries with the full context tool set, which can reopen investigation instead of converging.
Add a two-round terminal-only finalization phase after either condition. The model can only submit code_comment findings and finish with task_done, so gathered evidence is preserved without allowing more context expansion. Explicit successful task_done remains the only completion signal.
Automated Review Reproduction
The PR's OpenCodeReview check reproduced the existing behavior with the published v1.8.10 binary: reviewing only
internal/llmloop/loop.gomade 19 context-expansion calls (8code_search, 11file_read), consumed 175,414 tokens, and timed out after 10 minutes without callingtask_done.This check intentionally runs the trusted base-branch action and installs the published
latestpackage underpull_request_target; it does not execute the implementation from this fork. The regular CI, race tests, and cross-compilation checks pass on the PR code.Type of Change
How Has This Been Tested?
Additional verification:
Checklist
Related Issues
No linked issue.