fix: preserve reviewer verdict on salvage - #714
Merged
Merged
Conversation
Collaborator
Author
🔍 Fresh review of a3ddccc (
|
Merged
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Root cause
A real reviewer session for AX-1366 exhausted 30 tool turns. The generic no-tools salvage prompt only asked for the original task format, so the model returned a worker-style status summary with no
Decision/JSON. The strict reviewer parser correctly rejected it, producing the recurring AGT-4484 bucket.Fix
When salvage ends a review/verdict task, explicitly require
Decision: approve|revise|reject, concrete reasoning, and forbid a worker-status summary. This applies through the shared agentic loop used by all in-process adapters.Verification
npx vitest run src/adapters/agenticLoop.test.ts src/adapters/resultParsing.test.ts src/agents/reviewer.test.ts --reporter=dot— 139 passednpm run typechecknpm run lintgit diff --checkRegression test drives tool exhaustion and asserts the salvage prompt contains the explicit verdict contract.