Discover Claude models from the live harness - #21
Conversation
Bugbot is not enabled for your account, so this pull request was not reviewed. Enable Bugbot in the Cursor dashboard to get automatic reviews on future PRs. |
Warning Review limit reached
Next review available in:52 minutes Enable usage-based reviews in Billing to review now. Otherwise, wait until the next included review is available. How can I continue?After more reviews become available, a review can be triggered using the To avoid repeated limits, reduce automatic review volume by pausing incremental auto-reviews earlier, using label-based review opt-in, excluding WIP or generated PR titles, or requesting reviews manually when the PR is ready. If your team needs uninterrupted high-volume reviews, an organization admin can enable usage-based reviews. How do review limits work?CodeRabbit enforces per-developer PR review limits for each organization. Most developers receive the normal plan review availability. For paid Pro and Pro+ PR reviews, CodeRabbit uses adaptive limits for sustained high-volume activity. When a developer's recent PR review activity reaches the 95th percentile or higher among CodeRabbit users, additional reviews become available more gradually as earlier reviews age out of the rolling window. Please refer docs for additional details. Review details⚙️ Run configurationConfiguration used: Organization UI Review profile: ASSERTIVE Plan: Pro Plus Run ID: 📒 Files selected for processing (2)
📝 WalkthroughWalkthroughClaude model initialization metadata is parsed into provider models with effort and fast-mode capabilities. The capabilities probe exposes these models, and provider status uses them when available while retaining built-in fallbacks. ChangesClaude model discovery
Estimated code review effort: 4 (Complex) | ~45 minutes Sequence Diagram(s)sequenceDiagram
participant ClaudeSDK
participant ClaudeCapabilitiesProbe
participant ClaudeProvider
participant ProviderRegistry
ClaudeSDK->>ClaudeCapabilitiesProbe: return initialization models
ClaudeCapabilitiesProbe->>ClaudeProvider: expose parsed capabilities.models
ClaudeProvider->>ProviderRegistry: select discovered models
ProviderRegistry-->>ClaudeProvider: return provider snapshot
Suggested reviewers: Poem
🚥 Pre-merge checks | ✅ 5✅ Passed checks (5 passed)
✨ Finishing Touches🧪 Generate unit tests (beta)
Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out. Comment |
There was a problem hiding this comment.
💡 Codex Review
Here are some automated review suggestions for this pull request.
Reviewed commit:f63e8b3f29
ℹ️ About Codex in GitHub
Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you
- Open a pull request for review
- Mark a draft as ready
- Comment "@codex review".
If Codex has suggestions, it will comment; otherwise it will react with 👍.
Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".
Uh oh!
There was an error while loading. Please reload this page.
There was a problem hiding this comment.
💡 Codex Review
Here are some automated review suggestions for this pull request.
Reviewed commit:fdd344d4a2
ℹ️ About Codex in GitHub
Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you
- Open a pull request for review
- Mark a draft as ready
- Comment "@codex review".
If Codex has suggestions, it will comment; otherwise it will react with 👍.
Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".
| ? yield* resolveCapabilities(claudeSettings).pipe(Effect.orElseSucceed(() => undefined)) | ||
| : undefined; | ||
| const models = providerModelsFromSettings( | ||
| capabilities?.models.length ? capabilities.models : fallbackModels, |
There was a problem hiding this comment.
During normal startup this selects the account-specific catalog, but the refresh is merged into makePendingClaudeProvider's snapshot, which already contains every built-in model. mergeProviderSnapshot in ProviderRegistry.ts retains missing models for every non-OpenCode provider and appends them at lines 81-83 and 120-123, so unavailable built-ins—and cached models discovered under a previous Claude account—remain visible and selectable after a successful live probe. Successful Claude inventories need to replace, rather than merge with, the pending/cached model list.
Useful? React with 👍 / 👎.
Problem
Claude Code already reports the account-specific live model catalog during SDK initialization, but T3 discarded it and used only a hard-coded list. New Claude models therefore required a T3 code release before they could appear in the app.
Solution
Verification
vp test run apps/server/src/provider/Layers/ClaudeCapabilitiesProbe.test.ts apps/server/src/provider/Layers/ProviderRegistry.test.ts— 45 tests passed.vp run --filter t3 typecheck— passed.vp lintandvp fmt --check— passed.vp run --filter t3 build— passed.Impact
Backend provider discovery only; no schema, migration, auth, relay, or infrastructure changes. No screenshots are required because the existing model picker already renders provider snapshot models. After merge this will be published as one checksummed DevPC managed T3 artifact and pinned once in
remote-dev.Summary by CodeRabbit
New Features
Improvements