docs(tiers): make Opus 5 the default top tier, Fable the peak reserve - #32
Merged
Conversation
Claude Opus 5 reaches within ~0.5% of Fable 5's peak intelligence at roughly half the cost, so route the everyday top-tier work to Opus and keep Fable only for the last-0.5% max-effort cases. - CLAUDE.md: §2 tier split + 4-row tier table + when-in-doubt rule now put planning, all review loops, iteration, gnarly debugging, and non-trivial implementation on Opus (default); Sonnet keeps simpler decided-shape implementation; Fable is a peak reserve (~2x Opus cost). §1c local review loop reviews on Opus (Fable only for peak calls); §1 pre-commit review runs on Opus. Cross-model mapping updated. - subagent-strategy.md: PR-shipping split + model rubric mirror the same ladder; Fable bullet reframed as the deliberate peak escalation. - git-workflow.md: pre-commit review and ci/cr/merge-watch fix-push and exploratory-debug escalations run on Opus; Fable reserved for the hardest money-path diffs. - issue-pr-autopilot.md: added a note that frontier-judgement escalation to Fable is a manual, human-driven re-plan (single-model cloud routines cannot escalate mid-run), plus an operational caveat that the RemoteTrigger API pins the cron minute server-side.
Contributor
|
Caution Review failedThe pull request is closed. ℹ️ Recent review info⚙️ Run configurationConfiguration used: defaults Review profile: CHILL Plan: Pro Run ID: 📒 Files selected for processing (4)
📝 WalkthroughWalkthroughThe documentation updates add Fable to model-tier guidance, make Opus the standard review tier, require explicit model selection for Agent calls, define repository initialization and cleanup procedures, and clarify autopilot scheduling and manual escalation. ChangesAgent tier and review guidance
Estimated code review effort: 1 (Trivial) | ~5 minutes Possibly related PRs
Suggested labels: ✨ Finishing Touches🧪 Generate unit tests (beta)
Comment |
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.
Routes everyday top-tier work to Opus 5 (planning, all review loops, iteration, gnarly debugging, non-trivial implementation) and keeps Fable as a peak reserve used only when the last ~0.5% of max-effort intelligence decides the outcome. Per the Claude Opus 5 announcement: Opus 5 lands within ~0.5% of Fable 5's peak at roughly half the cost.
Files: CLAUDE.md (§2 split + tier table + §1c review loop + §1 pre-commit review + cross-model mapping), subagent-strategy.md (PR-shipping split + model rubric), git-workflow.md (review + watcher fix-push tiers), issue-pr-autopilot.md (manual Fable-escalation note + RemoteTrigger server-pinned-minute caveat).
User-directed change; not a self-noticed guidance gap.
Summary by CodeRabbit