docs: add OpenAI model tier mapping alongside Anthropic tiers#5
Conversation
CLAUDE.md may also be loaded by OpenAI tooling. Adds a mapping note near the Core Tenets so an OpenAI-driven session reads the same rules and resolves Anthropic tier names to the equivalent OpenAI tiers (Haiku → GPT-5.4-mini, Sonnet → GPT-5.4, Opus → GPT-5.5). Also extends the §2 model-rubric block with three parallel "OpenAI mini-tier / mid-tier / top-tier" bullets so an OpenAI-side reader sees the same task examples without having to mentally map the Anthropic ones, and generalises the "Default: delegate to the cheapest sufficient tier" + "When in doubt, go one tier cheaper" rules to cover both model families.
|
No actionable comments were generated in the recent review. 🎉 ℹ️ Recent review info⚙️ Run configurationConfiguration used: defaults Review profile: CHILL Plan: Pro Run ID: 📒 Files selected for processing (1)
📝 WalkthroughWalkthroughUpdates to CLAUDE.md: adds explicit OpenAI↔Anthropic tier mappings, modifies delegation rules to allow cheaper OpenAI subagents as fallback, expands the tier-to-task-complexity rubric with OpenAI mini/mid/top categories, and generalizes the cheaper-tier fallback across both model families. Changes
Estimated code review effort🎯 2 (Simple) | ⏱️ ~10 minutes Possibly related PRs
Poem
🚥 Pre-merge checks | ✅ 5✅ Passed checks (5 passed)
✏️ Tip: You can configure your own custom pre-merge checks in the settings. ✨ Finishing Touches🧪 Generate unit tests (beta)
Review rate limit: 2/5 reviews remaining, refill in 33 minutes and 1 second. Comment |
![coderabbitai[bot]](https://avatars.githubusercontent.com/in/347564?s=60&v=4){kind=small}
There was a problem hiding this comment.
Actionable comments posted: 2
🤖 Prompt for all review comments with AI agents
Verify each finding against the current code and only fix it if needed.
Inline comments:
In `@CLAUDE.md`:
- Around line 212-220: The "Haiku (default for most delegations)" phrasing can
be misread across model families; update the Haiku bullet and the "OpenAI
mini-tier" bullet so "default" is explicitly scoped per family or removed—e.g.,
change "Haiku (default for most delegations)" to "Haiku (default for Claude
family delegations)" or drop "default" and rely on the "cheapest sufficient
tier" rule, and mirror that clarification in the "OpenAI mini-tier (same role as
Haiku)" line so the default tier language is unambiguous between Claude and
OpenAI.
- Around line 17-23: Update the Anthropic→OpenAI mapping entries so the OpenAI
model identifiers use the lowercase "gpt-" prefix: change "GPT-5.4-mini" to
"gpt-5.4-mini", "GPT-5.4" to "gpt-5.4", and "GPT-5.5" to "gpt-5.5" in the
mapping block (the Haiku, Sonnet, and Opus lines in CLAUDE.md).
🪄 Autofix (Beta)
Fix all unresolved CodeRabbit comments on this PR:
- Push a commit to this branch (recommended)
- Create a new PR with the fixes
|
@coderabbitai review |
✅ Actions performedReview triggered.
|
There was a problem hiding this comment.
🧹 Nitpick comments (1)
CLAUDE.md (1)
217-220: 💤 Low value(Optional) Clean up the LanguageTool “repeated word” warning near the OpenAI top-tier bullet.
Static analysis reported three successive sentences beginning with the same word around the OpenAI rubric area (near Line 219). This won’t affect behavior, but if you’re keeping CLAUDE.md lint-clean, consider a small reword to remove the repeated sentence-start pattern.
🤖 Prompt for AI Agents
Verify each finding against the current code and only fix it if needed. In `@CLAUDE.md` around lines 217 - 220, Reword the three successive sentences that begin the OpenAI rubric bullets to avoid repeating the same sentence-start pattern near the "OpenAI top-tier" bullet: edit the bullet text for "OpenAI mini-tier", "OpenAI mid-tier", and/or "OpenAI top-tier" in CLAUDE.md so the sentences no longer all start with the same word (e.g., change one bullet's leading clause to "Includes" or "Handles" or rephrase the OpenAI top-tier sentence), preserving original meaning and examples while keeping the bullets consistent and lint-clean.
🤖 Prompt for all review comments with AI agents
Verify each finding against the current code and only fix it if needed.
Nitpick comments:
In `@CLAUDE.md`:
- Around line 217-220: Reword the three successive sentences that begin the
OpenAI rubric bullets to avoid repeating the same sentence-start pattern near
the "OpenAI top-tier" bullet: edit the bullet text for "OpenAI mini-tier",
"OpenAI mid-tier", and/or "OpenAI top-tier" in CLAUDE.md so the sentences no
longer all start with the same word (e.g., change one bullet's leading clause to
"Includes" or "Handles" or rephrase the OpenAI top-tier sentence), preserving
original meaning and examples while keeping the bullets consistent and
lint-clean.
cristim
added
triaged
|
@coderabbitai review |
✅ Actions performedReview triggered.
|
1 participant
Summary
CLAUDE.mdmay also be loaded by OpenAI tooling. Adds an explicit Anthropic→OpenAI tier mapping near the Core Tenets so an OpenAI-driven session resolves Anthropic tier names (Haiku / Sonnet / Opus) to the equivalent OpenAI tiers without having to guess:The mapping note also calls out that the cheapest/mid/top-tier alignment should be kept in sync if local OpenAI model names change.
§2 "Subagent Strategy" additionally gets three parallel OpenAI mini-tier / mid-tier / top-tier bullets in the model rubric so an OpenAI-side reader sees the same task examples directly, and the existing "Default: delegate to the cheapest sufficient tier" + "When in doubt, go one tier cheaper" rules are generalised to cover both model families instead of naming Haiku/Sonnet specifically.
Test plan
CLAUDE.mdlines 14–22 standalone — the OpenAI mapping is unambiguous and locatable from the Core TenetsCLAUDE.md§2 "Subagent Strategy" — the three new OpenAI bullets sit alongside the Haiku/Sonnet/Opus ones, and the "When in doubt" + "Default: delegate down" rules read consistently for both familiesgit diff --stat origin/mainshows exactly one file changed (CLAUDE.md), 12 insertions / 2 deletionsSummary by CodeRabbit