Sonnet floor for persona reviewers + honest review-preferences semantics - #15
Merged
Conversation
…eferences semantics Reviewer dispatch defaulted to haiku — judgment work on the cheapest tier produces shallow/malformed findings that cost more than the tier saves. New rule: orchestrator agent-model wins when defined; otherwise sonnet; haiku reserved for mechanical sub-tasks. ce-run Step 6 described review-preferences keys no orchestrator defines (min-reviewers, require-categories, prefer-categories) and told the model preferences 'add constraints on top of' ce:review's selection — one of the causes of double-spawned review rosters (observed 2026-08, 14+ concurrent agents). Now documents the real keys and the roster: replace substitution semantic. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_014ZyVXdjCUAwHbunnFoREUJ
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
1 participant
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Two fixes from the 2026-08-28 Erin runaway-spawning investigation (ce-reviewers-jsl PRs EveryInc#27/EveryInc#28 are the orchestrator side; this is the plugin side):
ce-review model tiering: persona reviewers dispatched on haiku by default — judgment work on the cheapest tier returns shallow or malformed findings that cost more than the tier saves. Now: an orchestrator's
agent-model:wins when defined; defaultsonnet; haiku reserved for mechanical sub-tasks.ce-run Step 6: documented
review-preferenceskeys that no orchestrator defines (min-reviewers,require-categories,prefer-categories), and instructed that preferences "add constraints on top of" ce:review's selection — a contributing cause of double-spawned rosters (14+ concurrent review agents). Now documents the keys orchestrators actually use and theroster: replacesubstitution semantic Erin relies on.No version bump or changelog entry per the repo's release-please policy.
🤖 Generated with Claude Code
https://claude.ai/code/session_014ZyVXdjCUAwHbunnFoREUJ