feat(frontend): OpenAI-compatible endpoint label and harness-gated picker#5346
Conversation
|
The latest updates on your projects. Learn more about Vercel for GitHub.
|
|
@coderabbitai review |
✅ Action performedReview finished.
|
|
Important Review skippedAuto reviews are disabled on base/target branches other than the default branch. Please check the settings in the CodeRabbit UI or the ⚙️ Run configurationConfiguration used: Organization UI Review profile: CHILL Plan: Pro Plus Run ID: You can disable this status message by setting the Use the checkbox below for a quick retry:
📝 WalkthroughWalkthroughThe PR renames the custom provider to “OpenAI-compatible endpoint” across the UI and tests, maps custom deployments to the OpenAI provider family, tightens capability gating, and adds unit coverage for filtering and model-reference composition. ChangesCustom provider flow
Estimated code review effort: 3 (Moderate) | ~20 minutes Possibly related PRs
🚥 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 |
Railway Preview Environment
Updated at 2026-07-17T22:55:35.950Z |
Release v0.105.4 assembly note — held, needs a one-file fixRetargeted this PR's base to Root cause — one stale shared fixture. Every one of the 12 failures dies in the same precondition helper: This PR renames/restructures the "Custom Provider" section label in Fix: update Note: base branch |
c27bc9d to
5b047bc
Compare
feat(frontend): OpenAI-compatible endpoint label and harness-gated picker
Context
The vault provider kind
customwas labeled "Custom provider" in the settings and connectionforms. The name did not say what the kind is: a single-key OpenAI-compatible Chat Completions
endpoint at a base URL you name. Selecting a custom connection also did not set a provider family,
so a bare model from a custom connection could carry over whatever provider was selected before it.
This change relabels the kind and makes the picker pick the right family. It pairs with the backend
change that lets these connections run on Pi.
Changes
The
customprovider-kind label now reads "OpenAI-compatible endpoint". The stored value isunchanged; only the display string moves.
Selecting a custom connection defaults the provider family to
openaiand preserves the connectionslug. Selecting an explicitly family-qualified model keeps that family. Switching from an unrelated
prior provider no longer leaks the prior provider into the new custom selection.
The picker is harness-gated. Custom vault models appear only where the harness reaches them. Pi
shows compatible custom connections when its custom deployment is enabled; Claude does not show an
OpenAI-compatible bare-model connection. Existing explicit Anthropic gateway behavior is unchanged.
Before:
PROVIDER_LABELS.customrendered "Custom provider", and a custom selection could submit thepreviously selected provider. After: it renders "OpenAI-compatible endpoint", and a bare-model
custom selection writes provider
openaiwith the exact slug.Scope / risk
custom.harness deployment set. This PR should merge after or with the backend PR
(
feat(agents): OpenAI-compatible endpoints for Pi via vault custom connections). Merged alone,the label changes but no new connection can run.
model-hub.tsstrings are updated to match the new label. A live Playwright run isstill outstanding (tracked in the project open-issues log).
How to QA
"OpenAI-compatible endpoint". The connection you already have is unchanged.
provider resolves to
openaiand the connection slug is preserved in the submitted config.does not carry the old provider.
bare-model connection does not appear; an existing Anthropic gateway still does.
Test command:
Evidence: entity-ui suite 208 passed. The manual UI click-through passed on the EE dev stack
(2026-07-14 and 2026-07-15).
https://claude.ai/code/session_01AYBi6cYuQCR87dAD3ZdPWc