When requests arrive via OpenAI-compatible endpoints and are converted to
Gemini format, functionCall parts were missing thoughtSignature. Gemini 3+
rejects such requests with 400 when thinking mode is enabled.
Changes:
- Extract thoughtSignature from tool_calls provider_specific_fields
- Inject into corresponding functionCall parts
- Fall back to dummy sentinel when thinking is enabled but no signature available
- Split signed/unsigned parts into separate model messages (Gemini requirement)
All existing tests pass (59/59).
- Add localCallbackServer for handling OAuth callback requests.
- Implement service functions for managing OAuth sessions and accounts.
- Create CodexAdapter for platform-specific interactions.
- Develop OAuthManagement page for managing OAuth connections and providers.
- Add tests for localCallbackServer and OAuthManagement components.
Thread account-level proxyUrl from extraConfig into
resolveChannelProxyUrl for endpointFlow and the third parameter
of withSiteRecordProxyRequestInit for direct fetch calls across
chat, responses, completions, embeddings, images, search, and
videos proxy routes.
The ExecuteEndpointFlowInput.proxyUrl field was defined but ignored
by the fetch call. Apply it via withExplicitProxyRequestInit when
present, falling back to withSiteProxyRequestInit otherwise.
- Added support for preserving native response.compaction payloads when compact serialization is requested in `serializeResponsesFinalPayload`.
- Implemented serialization of compact mode as `response.compaction` instead of an ordinary response object.
- Introduced a new type `ResponsesFinalSerializationMode` to manage serialization modes.
- Refactored `serializeResponsesFinalPayload` to handle different serialization modes and usage payloads.
- Added tests to ensure correct behavior for compact serialization and preservation of upstream payloads.
- Updated favicon assets and documentation configuration for VitePress.
- Added `routeCompatibility.ts` to handle response compatibility for different endpoints.
- Introduced `chatEndpointStrategy.ts` to manage chat endpoint recovery and downgrading logic.
- Created tests for chat formats and input file handling in `chatFormatsCore.test.ts` and `inputFile.test.ts`.
- Developed `endpointCompatibility.ts` to define compatibility checks and header building for various endpoints.
- Implemented protocol lifecycle management in `protocolLifecycle.ts` and corresponding tests in `protocolModel.test.ts`.
- Enhanced tooltip functionality with a new `TooltipLayer` component and associated tests.
- Refactored existing components to improve tooltip handling and ensure proper rendering in the application.
- Updated modelTesterSession tests to serialize assistant history as output_text and removed mime_type from input_file.
- Introduced new tests for stats proxy logs routes, covering paginated log retrieval and detailed log views with parsed billing details.
- Added server-driven tests for ProxyLogs component to validate API interactions and UI rendering based on server data.
- Updated chat stream tests to clarify reasoning behavior and ensure correct handling of include lists.
- Enhanced responses proxy route to normalize include lists and apply default reasoning includes based on request context.
- Improved transformer metadata handling, ensuring proper serialization and deserialization of metadata records.
- Added tests for preserving reasoning signatures and managing citations in OpenAI-compatible responses.
- Refined the logic for injecting default includes in responses and ensured compatibility with codex surface defaults.
- Added `inputFiles.ts` for handling input file resolution, including validation and normalization of file types.
- Introduced `proxyFileStore.ts` for managing file storage, retrieval, and deletion with owner-based isolation.
- Created tests for proxy file storage functionality in `proxyFileStore.test.ts`.
- Developed `proxyInputFileResolver.ts` to resolve input files in OpenAI and response formats.
- Added utility functions for handling reasoning signatures in `reasoningTransport.ts` and corresponding tests.
- Enhanced shared input file transformers in `inputFile.ts` for better file handling and type inference.
- Implement tests for anthropicMessagesTransformer including serialization and event handling in stream.test.ts.
- Add usage extraction tests for anthropicMessagesUsage in usage.test.ts.
- Create extensive tests for Gemini generate content aggregation and outbound usage in aggregator-outbound-usage.test.ts.
- Introduce tests for Gemini stream handling and payload parsing in stream.test.ts.
- Enhance OpenAI response serialization tests to ensure proper mapping of usage details and preservation of annotations in outbound.test.ts.
- Introduced `NormalizedContentBlock`, `NormalizedUsage`, and related types for structured data handling.
- Implemented utility functions for creating and merging normalized usage statistics.
- Added a new module for shared transformers to manage chat formats.
feat: Implement app local state management
- Created functions to clear application installation state and manage local storage keys.
- Added constants for theme and user profile storage keys.
test: Add tests for BrandIcon component rendering
- Implemented tests to ensure BrandIcon does not force a colored background for recognized brands.
test: Enhance ModelAnalysisPanel tests
- Added tests for rendering token summaries with compact units in ModelAnalysisPanel.
test: Add tests for ModernSelect component
- Implemented tests to verify icon rendering for selected options in ModernSelect.
test: Add number formatting tests
- Created tests for `formatCompactTokenMetric` to validate token formatting logic.
feat: Implement account creation intent handling
- Added logic to handle session and API key creation intents in the Accounts page.
test: Add tests for dashboard performance card
- Implemented tests to verify rendering of performance metrics in the dashboard.
feat: Implement default connection segment resolution
- Added logic to determine initial connection segments based on platform types.
test: Add tests for settings factory reset functionality
- Implemented tests to verify the factory reset process and local state clearing.
test: Add tests for site creation redirect logic
- Implemented tests to verify redirection based on site capabilities during creation.
test: Add tests for token routes refresh decision action
- Implemented tests to verify refresh decision actions in TokenRoutes.
test: Add tests for token routes snapshot load
- Implemented tests to verify loading of cached snapshot data in TokenRoutes.
- Added editing capabilities for route management in TokenRoutes, allowing users to modify existing routes.
- Introduced a reset function for the route form to clear inputs after adding or editing routes.
- Implemented a visibility filter for routes, ensuring exact routes covered by group routes are hidden from the main list.
- Updated tests to cover new functionalities, including editing routes and verifying visibility logic.
- Created a new helper function to build a list of visible routes based on defined criteria.
- Enhanced schema compatibility checks for site configurations, ensuring necessary columns are added to the database.
- Implement tests for the CenteredModal component to ensure it follows the shared modal shell pattern.
- Add comprehensive tests for the SearchModal results, including API interactions and UI behavior.
- Create tests for accounts batch actions, verifying the functionality of refreshing balances and selecting accounts via row clicks.
- Introduce tests for the accounts centered modal to confirm the use of CenteredModal for add/edit/rebind flows.
- Add tests for the accounts edit panel to ensure proper rendering and functionality when opening the edit modal.
- Implement tests for the accounts tokens header to verify the reuse of the accounts page header in the tokens segment.
- Create tests for the system proxy settings, ensuring the correct saving of the proxy URL.
- Add tests for the tokens actions layout to confirm the proper layout and styling of token row actions.
- Implement tests for the tokens batch actions, including deletion of selected tokens and row selection behavior.
- Introduce tests for the tokens focus navigation to ensure the correct highlighting of focused token rows.
- Removed apiKey from SiteRow type and related forms in Sites component.
- Updated site saving logic to handle site creation without apiKey.
- Adjusted tests to ensure apiKey is not included in site editor state.
- Refactored Tokens component to support embedded mode and redirect logic.
- Added migration tests for site apiKey to account migration service.
- Implemented siteApiKeyMigrationService to handle migration of apiKey to account tokens.
- Enhanced tests for segmented connections in Accounts and redirect logic for Tokens.
Some upstream providers (e.g. certain OpenAI-compatible aggregators)
emit thinking/reasoning tokens with the same text written to both
delta.content and delta.reasoning_content simultaneously.
When metapi forwarded these chunks as-is, downstream clients would read
delta.content and display the internal reasoning text to end users —
appearing as word-by-word "thinking narration" in chat interfaces.
Fix: after resolving both contentDelta and reasoningDelta, suppress
contentDelta when it is non-empty and identical to reasoningDelta.
This covers the duplicate-write pattern without affecting any legitimate
case (a provider would never intentionally set content == reasoning).
Hureru