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.
- Added support for specifying prefix lengths for live text-backed columns in MySQL upgrade SQL generation.
- Introduced `mysqlIndexPrefixRequirements` option to `generateUpgradeSql` function.
- Updated tests to validate the new functionality for MySQL upgrade generation.
- Removed legacy shared index compatibility specifications to avoid duplication with runtime schema bootstrap.
- Improved error handling and response structure in account token coverage refresh logic.
- Added tests for coverage refresh failure scenarios to ensure proper handling of token persistence.
# Conflicts:
# src/server/db/sharedIndexSchemaCompatibility.test.ts
# src/server/db/sharedIndexSchemaCompatibility.ts
# src/server/index.ts
- Implemented a DeleteConfirmModal component for confirming deletions across various pages (Accounts, Sites, Tokens).
- Updated Accounts and Sites pages to utilize the new modal for single and batch deletions.
- Enhanced TokensPanel to support group selection and improved token deletion handling with confirmation.
- Added functionality to skip confirmation for batch actions and handle token group fetching with error management.
- Improved error handling for expired sessions in NewApiAdapter and OneApiAdapter.
- Added tests for new delete confirmation behavior and token editing interactions.
- fix(api): use effectiveTokenId in route channel insert to match duplicate detection logic
- fix(css): rename keyframes to kebab-case (filter-bar-slide-down/up)
- fix(web): wrap "保存中..." text with tr() in ManualRoutePanel
- fix(web): use ref for fresh cache check and add error handling in useRouteChannels
Add GET /api/routes/summary for lightweight first-screen data,
GET /api/routes/:id/channels for on-demand channel loading,
and POST /api/routes/:id/channels/batch for bulk channel creation.
Extract fetchChannelsForRoutes() as reusable helper.
- 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.
cita