- Updated tests for chat model provider registry to include providerKind in configurations.
- Improved human memory renderers tests for better readability and accuracy in action definitions.
- Translated cancellation message in LLM executor text tests to Chinese.
- Revised OpenAI compatible runtime tests to ensure explicit provider kind is retained.
- Added new skill definitions in runtime skill contract tests.
- Enhanced stagehand model config tests to clarify provider mappings.
- Updated conversation composer tests to validate AI preference selections and workspace interactions.
- Refined session creation tests to ensure proper title updates and help button visibility.
- Improved run blocker tests to validate stopping long tasks and approval runs.
- Enhanced chat run service contract tests to include preference settings.
- Added image recognition tool harness for better testing capabilities.
- Updated real AI dependencies factory to include strategy preference storage.
- Improved workbench utility functions for better interaction with UI elements.
- Introduced LlmStrategyPreferenceStore to handle user preferences for LLM strategies.
- Updated StorageService to include the new LlmStrategyPreferenceStore.
- Modified LlmConfigStore to validate provider kinds using isLlmProviderKind.
- Enhanced SQLite migration scripts to create llm_strategy_preference table.
- Updated structured output handling to support new strategy preferences in various components.
- Refactored tests to accommodate changes in structured output strategy handling.
- Added mergeById method in ChatStore for upsert operations on chat sessions and messages.
- Refactored saveChatSession and saveChatMessage methods to utilize mergeById.
- Introduced getChatMessagesPage method for paginated retrieval of chat messages.
- Enhanced getChatSession to return session details and handle non-existent sessions.
- Updated DocumentsStore to ensure type safety with unknown records.
- Expanded types in store.d.ts for new chat session requests and responses.
- Updated preload API to include new chat session methods.
- Optimized MarkdownRenderer and AssistantTurnCard components with React.memo for performance.
- Enhanced WorkbenchComposer to display hints based on submission state.
- Improved ChatPage to manage session switching and load older messages efficiently.
- Added performance tests for chat session open operations.
- Updated e2e tests to cover new session management features and ensure reliability.
- Implement tests for workspace index job store contract to validate job creation, active querying, phase progress updates, and failure convergence.
- Add tests for interrupted job recovery to ensure that running jobs and file indexing statuses are marked as failed upon recovery initiation.
- Enhance workspace prepare contract tests to ensure that active jobs are reused instead of creating new ones during indexing.
- Introduce end-to-end tests for restoring progress in workspace preparation and scanning, ensuring UI reflects persistent job states and disabled buttons during operations.
- Update web-related tests to validate the summary refresh behavior after web source indexing and scanning, ensuring accurate status updates in the UI.
hy