The Zod updateSettingsSchema was missing fallbackStrategy,
wildcardAliases, and stickyRoundRobinLimit fields. Since
.passthrough() was removed in a previous cleanup, these unknown
keys were silently stripped during validation, so the PATCH
/api/settings call never actually persisted these values.
Wrap fs.mkdirSync(DATA_DIR) in try/catch so OmniRoute doesn't crash
when the user's home directory is not writable (e.g. restricted
environments, npm global install with different user).
Prints a clear warning with DATA_DIR env var recommendation.
- New 'Provider Proxy' button next to Connections heading
- Opens ProxyConfigModal at provider level
- Amber highlight when proxy is configured, shows IP
- Supports all 3 proxy levels: global, provider, per-connection
- Green: global proxy
- Yellow/amber: provider-level proxy
- Blue: per-connection proxy
- Always shows proxy host/IP in the badge text
- Tooltip shows proxy source and full address
- hasProxy now checks keys/providers/global config levels
- Badge color changed to emerald green to match other status badges
- Tooltip shows proxy source: per-connection, per-provider, or global
- Previously only showed for per-connection proxy, missing global/provider
Server-side: wrap getCliRuntimeStatus() in 5s Promise.race timeout
Client-side: add 8s AbortController timeout to fetchToolStatuses()
Prevents entire page from staying in skeleton state forever when
binary checks hang on VPS
- instrumentation.ts: use eval('require')('crypto') to bypass webpack
- HomePageClient.tsx: add useTranslations to ProviderOverviewCard and
ProviderModelsModal (separate components need own hooks)
- next.config.mjs: temporarily allow TS errors during build (remaining
sub-component scope issues in EvalsTab.tsx)
Use eval('require')('crypto') to hide the import from webpack's
static analysis. The instrumentation file is compiled for both
client and server, but crypto is only used in Node.js runtime.
Root cause: createProviderConnection() upsert logic for codex deduplicates
by workspaceId only (chatgpt_account_id from JWT). Two Pro Business accounts
from the same Team workspace resolve to the same workspaceId, causing the
second to silently overwrite the first.
Fix: compound uniqueness check using workspaceId + email, so different users
within the same workspace create separate connections. Includes backward-compat
fallback for old connections without email (they'll still be updated).
- Removed duplicate providers/new keys (selectProvider, apiKeyRequired, etc.)
that overlapped with pre-existing keys from earlier migration
- Removed duplicate backToProviders in both files
- Removed duplicate loadingPricing and databaseSize in settings namespace
- Updated providers/new/page.tsx to use original key names
- Verified 0 duplicates remain via Python namespace scan
- EndpointPageClient: full migration including cloud proxy modals,
endpoint sections (chat, embeddings, images, rerank, audio, moderations),
enable/disable cloud flows, sync steps, and status messages
- Expanded endpoint namespace from 10 to 55 keys in en.json and pt-BR.json
- CombosPage, ComboCard, ComboFormModal, TestResultsView: full migration
- Headers, model tags, metrics, strategy descriptions, validation
- Form labels, advanced settings, drag-and-drop model list, actions
- Expanded combos namespace from 19 to 65 keys in en.json and pt-BR.json
- ApiManagerPageClient: full migration including stats cards, table headers,
modals, permissions system, validation messages, and usage tips
- PermissionsModal: search, model selection, allow/restrict toggles
- Expanded apiManager namespace from 18 to 67 keys in en.json and pt-BR.json
- Added ApiKeyMetadata interface to replace 'any' types in apiKeyPolicy.ts
- Added error logging in catch blocks for getApiKeyMetadata() and checkBudget()
- Fixed claude-sonnet-4-6-thinking → claude-sonnet-4-6 mismatch in usage.ts importantModels
Follow-up fixes for merged PRs #131 and #128
isModelAllowedForKey() existed in src/lib/db/apiKeys.ts but was never
called anywhere. API keys with allowedModels restrictions could access
any model through any endpoint.
Changes:
- Add shared enforceApiKeyPolicy() middleware (model restriction + budget)
- Wire it into chat handler (replacing inline budget-only check)
- Wire it into all /v1/* endpoints: embeddings, images/generations,
audio/speech, audio/transcriptions, moderations, rerank
- Wire it into provider-specific endpoints: /v1/providers/[provider]/embeddings,
/v1/providers/[provider]/images/generations
The middleware checks:
1. Model restriction — if key has allowedModels, verify the model is permitted
2. Budget limit — if key has budget configured, verify it hasn't been exceeded
Fixes#130
- Combo cards now resolve UUID-based provider IDs to user-defined names using providerNodes
- API key names shown in full instead of masked (Op***w → OpenAI-Prod-Key); key IDs remain masked
- Logs/Analytics: OAI-compatible and ANT-compatible providers now show user-defined names from providerNodes instead of generic OAI-COMPAT labels
- All views (combo cards, combos form modal, logs table, quick filters, dropdowns, analytics) updated consistently
