CherryHQ/cherry-studio · error · Error

OVMS provider requires a non-empty `baseURL`. An empty value

Error message

OVMS provider requires a non-empty `baseURL`. An empty value would resolve fetch paths against the renderer process origin (app://, file://) and surface as opaque "Failed to fetch" errors.

What it means

A fail-fast guard inside createOvmsProvider(): if settings.baseURL is empty, the provider refuses to construct. An empty baseURL would make the OpenAI-compatible SDK resolve fetch paths against the Electron renderer origin (app:// or file://), producing opaque 'Failed to fetch' errors that hide the real cause. Throwing at construction converts that into a clear, actionable message.

Source

Thrown at src/main/ai/provider/custom/ovms/ovmsProvider.ts:42

export interface OvmsProvider extends ProviderV3 {
  (modelId: string): LanguageModelV3
  languageModel(modelId: string): LanguageModelV3
  embeddingModel(modelId: string): EmbeddingModelV3
  imageModel(modelId: string): ImageModelV3
}

/**
 * Unified OVMS provider — chat, embedding, and image off one `ProviderV3`,
 * mirroring `newapi-provider.ts`. OVMS is a local OpenVINO Model Server with
 * NO auth, so headers carry only what the caller passes (no `Authorization`).
 * Chat/embedding hit `settings.baseURL`; the image model keeps its bespoke
 * single-shot behavior via `createImageGenerationModel + createOvmsTransport`
 * aimed at `settings.imageBaseURL`.
 */
export function createOvmsProvider(settings: OvmsProviderSettings = {}): OvmsProvider {
  const { baseURL, fetch: customFetch } = settings
  if (!baseURL) {
    throw new Error(
      'OVMS provider requires a non-empty `baseURL`. An empty value would resolve fetch paths against the renderer process origin (app://, file://) and surface as opaque "Failed to fetch" errors.'
    )
  }

  const authHeaders = () => ({ ...settings.headers })

  const url = ({ path }: { path: string; modelId: string }) => `${withoutTrailingSlash(baseURL)}${path}`

  const createChatModel = (modelId: string) =>
    new OpenAICompatibleChatLanguageModel(modelId, {
      provider: `${OVMS_PROVIDER_NAME}.chat`,
      url,
      headers: authHeaders,
      fetch: customFetch
    })

  const transport = createOvmsTransport({
    baseURL: settings.imageBaseURL || DEFAULT_OVMS_BASE_URL

View on GitHub (pinned to 726446b54c)

Solutions

1. Set a non-empty OVMS host before constructing the provider (e.g. http://localhost:8000/v3/ for chat/embedding).
2. If constructing programmatically, read the host from BootConfig/Preference before calling createOvmsProvider.
3. In the UI, mark the baseURL field required and disable 'Save' until filled.
4. If the user genuinely has no host yet, do not instantiate the provider — defer construction until settings are complete.

Example fix

// before
const provider = createOvmsProvider({ baseURL: settings.ovms?.baseURL })
// after — validate before construction
if (!settings.ovms?.baseURL) throw new Error('OVMS host is required')
const provider = createOvmsProvider({ baseURL: settings.ovms.baseURL })

if (!settings.baseURL) {
  throw new Error('OVMS host is required (e.g. http://localhost:8000/v3/)')
}
const provider = createOvmsProvider(settings)

export function isOvmsBaseURLError(e: unknown): boolean {
  return e instanceof Error && /OVMS provider requires a non-empty `baseURL`/.test(e.message)
}

try {
  createOvmsProvider(settings)
} catch (e) {
  if (isOvmsBaseURLError(e)) {
    // prompt the user for the OVMS host before retrying
  }
  throw e
}

When it happens

Trigger: createOvmsProvider({}) or createOvmsProvider({ baseURL: '' }) — i.e. the OVMS provider is instantiated without a host. Happens at provider factory time when settings haven't been loaded or the user added an OVMS provider without filling in the host.

Common situations: New OVMS provider added via UI but the host field left blank, settings migration leaving baseURL undefined, or programmatic provider construction without the host.

Related errors

• PPIO provider requires a non-empty `baseURL`. An empty value
• Either baseUrl or command must be provided
• DashScope provider requires a non-empty `baseURL`. An empty
• DMXAPI provider requires a non-empty `baseURL` to build the
• DMXAPI provider requires a non-empty `baseURL`. An empty val

AI-assisted analysis of CherryHQ/cherry-studio@726446b54c (2026-08-12). Data as JSON: /api/errors/a675e651120d09ac. Report an issue: GitHub.