- Updated ModelSource type to include 'huggingface'.
- Enhanced MODEL_SOURCE_LABELS for better clarity.
- Introduced defaultModelSource function to determine default source based on service route.
- Modified AppSettings to include language and serviceRoute.
- Created end-to-end tests for language and service route settings.
- Added unit tests for application language and service preferences.
- Updated free models test to reflect new catalog naming conventions.
- Enhanced official API tests to ensure correct handling of service regions.
- Adjusted release artifact tests to accommodate internationalization.
- Updated SQLite migration and storage tests to include new preferences.
- Changed `requesting` to `streaming` in `SubagentTaskView` type.
- Updated logic in `reduceAssistantProcessGroups` to handle streaming state and message timing.
- Modified `AiRunEventPayload` to reflect the new streaming property.
- Enhanced tests for human subagent to validate streaming behavior and prefix cache retention.
- Added new test case for subagent prefix cache retention during live execution.
- Introduced assertions for prefix cache reuse in the continuation logic.
- Added a new DOCX fixture for testing document selection scenarios.
- Added new test cases for AI user question interactions, including scenarios for rejecting mixed tools and ensuring no auto-answering occurs.
- Updated the test plan documentation to include new test cases and clarify AI question handling processes.
- Improved the Workbench UI to support better user interactions with AI questions, including custom input handling and question navigation.
- Refactored backend code to implement a question resolver that manages AI question interactions and ensures proper event handling.
- Enhanced unit tests to validate the structure and behavior of the ask_user_question tool contract.
- Introduced end-to-end tests for the Workbench AI question flow, ensuring that user interactions are correctly captured and processed.
- Updated test plan documentation to reflect the new naming convention for prefix cache tests.
- Added new E2E test for prefix cache retention in the Workbench UI.
- Removed obsolete human-prefix-cache-retain-live test file and integrated its logic into the new test structure.
- Adjusted release preparation script to accommodate the new test case name.
- Enhanced LLM IO record handling in the E2E testing framework to support the new test requirements.
- Updated various harness and utility functions to ensure compatibility with the new test structure and naming.
- Updated multiple test files to remove unused parameters in page.evaluate calls.
- Simplified function signatures in waitForPreparedWebSources and other helper functions.
- Cleaned up assertions and formatting for better clarity in unit tests.
- Removed redundant code and improved consistency across test cases.
- Removed `test:ai`, `test:settings`, and `test:workbench` scripts from package.json.
- Updated test plan documentation to reflect the removal of `test:workbench`.
- Enhanced argument parsing in release scripts to ensure required values are provided.
- Added validation for unknown arguments in release scripts.
- Improved error handling for invalid arguments in test workflow.
- Updated unit tests to cover new validation logic for release arguments.
- Implemented `ppt-management-briefing-edit.showcase.test.ts` to test editing PPTs with AI assistance, including cost computation and screenshot capture.
- Created `ppt-management-briefing-review.showcase.test.ts` for reviewing PPTs, ensuring the generation of review screenshots and cost analysis.
- Added shared utility functions in `shared.ts` for handling showcase fixtures, workspace seeding, and cost computation.
- Developed `xinghe-kb-reply.showcase.test.ts` to validate knowledge base interactions, capturing screenshots of responses and ensuring correct tool usage.
- Introduced `xinghe-service-contract.showcase.test.ts` to test PDF contract inquiries, verifying unchanged source documents and capturing relevant screenshots.
- Configured Vitest for showcase tests in `vitest.showcase.config.mts`, setting appropriate timeouts and test inclusion/exclusion criteria.
- Added `showcase` as an explicit collection task in the testing framework, distinct from product regression.
- Introduced explicit collection task rules and updated documentation to reflect changes in testing structure.
- Updated E2E testing commands to utilize `--execution-profile` for better control over test execution.
- Implemented validation to prevent the selection of certain groups (e.g., `showcase`) during release command parsing.
- Enhanced regression selection logic to allow opt-in groups for release impact assessment while rejecting disallowed groups.
- Updated related scripts and test cases to ensure compliance with new testing rules and execution profiles.
- Introduced `design.md` to outline the constraints, storage layout, CDN/public channel, and local directory conventions for desktop releases.
- Added `release.md` detailing the release process for Windows and macOS, including tagging, candidate package creation, and COS publishing.
- Created `tencent-cdn-billing.md` to summarize billing and procurement for Tencent Cloud CDN and COS for domestic releases.
- Updated test plans in `ai-test.md` and `workbench-test.md` to reflect changes in release processes and added new tests for prefix cache retention.
- Enhanced release preparation script to assert live key configuration and run necessary live gate tests.
- Implemented a new test case for validating prefix cache retention during live interactions.
- Added contextWindowTokens to LlmPresetModel and updated mergePresetCatalogProviders to parse context window tokens from models.
- Removed compactionTriggerTokens from BrainContextBudgetSnapshot as it is no longer needed.
- Updated tests to reflect changes in context window calculations and compaction logic.
- Introduced new utility functions for resolving effective context window tokens and compaction tail tokens.
- Adjusted various test cases to ensure compatibility with the new context window handling.
- Improved error handling and logging in browser-related services and tests.
Upload and finalize artifacts under file.name (馆长-…) so CDN/URL
basename matches the save-as name without Content-Disposition.
storageName stays ASCII for on-disk resolve; update metadata maps
either alias to the public key.
COS CDN can keep immutable Content-Disposition after metadata updates
when purge is unavailable; trust origin head-object and warn if public
headers lag.
Use 馆长-${version}-macos-arm64.dmg as the user-visible installer name
while keeping ASCII storage keys, resolve on-disk assets by storageName
when they differ, and emit Windows-style UTF-8 Content-Disposition.
pnpm run now passes a literal -- into workflow.mjs, which rejects unknown args and blocks release:mac --test-only.
Co-authored-by: Cursor <cursoragent@cursor.com>
Failed pre-publish tags can be deleted and reused instead of stacking versions; docs and prepare errors point at release:abort.
Co-authored-by: Cursor <cursoragent@cursor.com>
- Removed support for macos-x64 in fragment definitions and related scripts.
- Updated finalize-macos script to only handle macos-arm64, simplifying signature checks and metadata writing.
- Deleted merge-mac-metadata script as it is no longer needed.
- Adjusted storage layout and release status checks to reflect the removal of x64.
- Updated tests to ensure they only validate macos-arm64 functionality.
- Removed bundled Playwright browser assertions and adjusted driver bundle copying to ignore local browsers.
- Introduced new assertions to ensure no Playwright browser files are bundled with the sidecar.
- Updated macOS and Windows release scripts to remove references to the sidecar browser cache.
- Enhanced WebCrawlerService to utilize an isolated Electron CDP connection for better browser context management.
- Added ElectronWebCrawlerBrowserHost class to manage browser connections and views.
- Updated tests to reflect changes in the web crawler service and ensure proper handling of CDP connections.
- Improved message rendering for AI interactions, including better formatting for tool calls and AI messages.
- Adjusted various test cases to align with new functionality and ensure robustness.
- Changed external links in settings-test.md and package.json to point to the new domestic website (guanzhangai.cn).
- Updated electron-builder.yml to publish updates from the new domestic URL.
- Introduced a new storage client for Tencent COS and refactored existing R2 client code to support both storage options.
- Implemented environment variable checks and configurations for Tencent COS and Cloudflare R2.
- Added tests for release storage configuration and ensured proper handling of live publish locks.
- Updated telemetry service to use the new official website URL for event reporting.
- Added a new UpdateRequiredPage component to handle scenarios where an application update is required before accessing the main interface.
- Integrated app startup status retrieval in the App component to determine if the application is in a normal state or requires an update.
- Updated routing logic to conditionally render the UpdateRequiredPage or the main application layout based on the startup status.
- Enhanced the AppUpdateContext to manage update actions and states effectively.
- Introduced new selectors for testing the update required page and its components.
- Added constants for app startup channels to facilitate communication between the main and renderer processes.
- Updated shared types to include AppStartupStatus for better type safety.
- Created unit tests for SQLite migration to ensure database integrity during updates.
- Implemented end-to-end tests to verify the update flow and recovery page functionality.
The test stage builds a clean workspace from the tag but only ran
pnpm install without seeding the prebuilt native cache, so package E2E
re-downloaded better-sqlite3 from GitHub release assets on every run and
failed on flaky network. Seed the cache like the prepare stage does.
- Set execution mode to 'exclusive' for specific e2e tests to ensure isolated execution.
- Set execution mode to 'live' for various e2e tests to enable real-time interactions.
- Refactor workspace binding in tests to use direct binding method for improved reliability.
- Enable file parallelism in Vitest configurations to enhance test execution efficiency.
- Remove obsolete end-to-end test for system notification session navigation.
- Enhance workspace scan progress tests to verify UI state restoration.
- Update web prepare progress tests to ensure correct UI behavior after navigation.
- Refactor unit tests for AI persistence and model configuration to improve naming consistency.
- Modify release tests to use localized product names and ensure correct artifact naming.
- Adjust vector and workspace backend tests to reflect unit test categorization.
- Improve error handling in web crawler test utilities.
- Update Electron app launch environment to support additional environment variables.
- Refactor end-to-end test cases for workbench to improve readability and maintainability.
- Introduced a new documentation file for the Tidy process, outlining steps for aligning documentation with implementation, organizing PRD and test cases, and verifying settings.
- Updated package.json to include a new test command for non-live AI tests and adjusted existing test commands for settings and updates.
- Implemented a cleanup function for local test logs in the release scripts to ensure a clean environment.
- Enhanced the release preparation script to include unit and non-live AI tests in the workflow.
- Added a new script to validate test architecture, ensuring that test cases align with defined plans and rules.
- Refactored existing test cases to improve organization and clarity, including renaming and restructuring tests for better maintainability.
- Adjusted timeout settings for specific tests to improve reliability.
- Corrected domain assignments in several unit tests to ensure proper categorization.
hy