Main process — worker thread for team data:
- New team-data-worker thread handles getTeamData and findLogsForTask,
isolating heavy file I/O (scanning 300+ subagent JSONL files) from
Electron's main event loop. getTeamData dropped from ~2000ms on the
main thread to ~110ms via the worker.
- Worker-side dedup and 10s result cache for findLogsForTask prevents
redundant scans when the same task is queried multiple times.
- Discovery cache TTL raised from 5s to 30s — avoids re-scanning the
entire project directory on every call.
- Message cap at 200 in TeamDataService to keep IPC payloads under 1MB
(was sending 2200+ messages / ~3MB, stalling Chromium IPC serialization).
- IPC handlers fall back to main-thread execution if the worker is
unavailable (graceful degradation).
Renderer — useShallow and memoization (55 files):
- Added useShallow to store selectors across 55 renderer files. Batched
individual useStore() calls (e.g. 17 calls in ExtensionStoreView,
10 in ConnectionSection) into single useShallow selectors, cutting
unnecessary re-render checks on every store update.
- MemberLogsTab: three 5-second polling intervals now pause when the
parent tab is hidden (display:none). Previously 5 hidden tabs × 3
intervals = 15 polling timers firing continuously.
- KanbanColumn wrapped in React.memo to skip re-renders when props
haven't changed.
- MemberList: memoized activeMembers/removedMembers/colorMap; replaced
O(n×m) per-member task scan with a pre-computed reviewer map.
- Bounded timer Maps in store initialization to prevent unbounded growth
of debounce/throttle tracking maps during long sessions.
- Added visual differentiation for 'task_comment' particles, adjusting size and glow effects.
- Updated drawing functions to handle new particle kind, ensuring proper rendering in the graph.
- Introduced a merge function for particles to prevent duplicates during state updates.
- Enhanced color constants for better visual representation of different particle types.
Root cause: handleTeammatePermissionRequest was called from 3 paths
(early inbox scan, Category 4 relay scan, stdout/native) but only
the early scan checked processedPermissionRequestIds, causing
duplicate ToolApprovalRequests and extra permission_responses.
Fix:
- Move processedPermissionRequestIds check INTO handleTeammatePermissionRequest
so ALL callers are protected by the same dedup gate
- Remove duplicate Category 4 scan that re-processed inbox messages
(early scan already covers all messages including read=true)
- Category 4 now only builds a filter Set to exclude permission_request
from relay to lead
- Added support for tracking task change presence with new IPC channels: TEAM_GET_TASK_CHANGE_PRESENCE and TEAM_SET_CHANGE_PRESENCE_TRACKING.
- Introduced JsonTaskChangePresenceRepository and TeamLogSourceTracker to manage task change presence data.
- Enhanced ChangeExtractorService to utilize task change presence services for improved task change detection.
- Updated TeamDataService to integrate task change presence tracking and resolve task change presence states.
- Modified UI components to reflect task change presence status in Kanban and task detail views.
This feature aims to provide real-time insights into task changes, enhancing user experience and task management capabilities.
- Replaced instances of isLeadAgentType with isLeadMember for improved clarity in team member role checks.
- Updated README to include a new built-in review workflow feature for agent task reviews.
- Enhanced team detail and list views to accurately reflect current team members and their roles.
- Improved CSS for step progress indicators, adding new styles for a circular stepper.
- Refactored provisioning progress block to utilize a new StepProgressBar component for better visual representation of steps.
- Updated `lookupMessage` function to enforce exact messageId matching, rejecting ambiguous cases.
- Improved documentation for `task_create_from_message` to clarify its requirements and usage.
- Enhanced `buildAssignmentMessage` and related functions to include messageId in task notifications for better provenance tracking.
- Added tests to validate the integration of messageId in task creation and relay processes, ensuring accurate message handling.
- Enhanced regex patterns in TeamMemberLogsFinder to allow for escaped sequences and increased character limits for text and content previews.
- Updated the MemberLogsTab component to prioritize recent previews over fallback mechanisms, ensuring more relevant message displays for lead users.
- Adjusted the length limit for returned previews to 1500 characters for better context in log summaries.
- Added support for tracking and displaying recent thinking previews alongside output previews in TeamMemberLogsFinder.
- Updated the return structure of log summaries to include last thinking previews and recent previews for better context.
- Enhanced the MemberLogsTab component to utilize recent previews for lead users, improving the clarity and relevance of displayed messages.
- Implemented filtering logic for task-scoped recent previews, ensuring only relevant messages are shown based on task intervals.
- Removed `remark-stringify` from the project to avoid ESM→CJS interop issues in Electron's main process.
- Updated the text formatting pipeline to use a custom plain-text compiler instead of `remark-stringify`.
- Enhanced task ID matching logic in `TeamMemberLogsFinder` to accommodate both full UUIDs and their short display forms, improving flexibility in task identification.
- Added comments to clarify the changes in task ID handling and regex construction.
- Added `remark-stringify` to the project for converting markdown to plain text.
- Updated the text formatting pipeline to include `remark-stringify` for improved markdown handling.
- Commented out performance logging in `TeamMemberLogsFinder` to reduce console output during execution.
- Enhanced the `getUnreadCount` and `getLegacyCutoff` functions in `commentReadStorage` to clarify legacy comment handling logic.
- Introduced a new function `quoteMarkdown` to format task comments with markdown quotes for better readability.
- Updated the `buildCommentNotificationMessage` to utilize `quoteMarkdown`, ensuring comments are displayed correctly.
- Refactored member role detection across multiple services to use `isLeadMember` for consistency and clarity in identifying team leads.
- Enhanced various components to improve handling of team member roles, ensuring accurate representation in UI and logic.
- Adjusted tests to reflect changes in comment formatting and member role checks, improving overall reliability.
- Added a new function to drop CLI provisioner members from the member map, ensuring these internal artifacts are never displayed to users.
- Updated the createCliProvisionerNameGuard function to unconditionally hide provisioner names, regardless of the presence of base members.
- Modified unit tests to reflect the new behavior of dropping provisioner names even when the base member is absent.
- Implemented deduplication logic in TeamMemberLogsFinder to retain only the largest cumulative subagent snapshots per sessionId and memberName.
- Enhanced MemberLogsTab with a new function to filter chunks based on task work intervals, improving the relevance of displayed logs.
- Updated state management in MemberLogsTab to utilize the new filtering logic for both preview and detail chunks.
- Added unit tests to verify the new deduplication and filtering functionalities.
- Modified comment notification formatting in tasks.js to enhance clarity by using bold text for task references.
- Updated the automated comment notification detection in maintenance.js to align with the new formatting.
- Adjusted test cases to reflect the updated comment structure for consistency in notifications.
- Increased FILE_MENTIONS_CACHE_MAX from 1000 to 10,000 to accommodate larger datasets.
- Introduced DISCOVERY_CACHE_TTL to optimize project session discovery by caching results.
- Updated findMemberLogs method to accept an optional mtimeSinceMs parameter for filtering logs based on modification time.
- Added lastOutputPreview to MemberLogSummary for displaying the last assistant output in logs.
- Implemented displayMemberName utility function to standardize member name display across various components.
- Updated multiple components to utilize displayMemberName for consistent member name rendering.
- Introduced a new method to collect subagent file candidates, streamlining the file scanning process.
- Implemented parallel processing of subagent files with a concurrency limit to enhance performance.
- Improved error handling during file parsing, ensuring that failures do not disrupt the overall scanning process.
- Added unit tests to verify that member logs are returned sorted by start time in descending order.
- Increased FILE_MENTIONS_CACHE_MAX from 200 to 1000 to accommodate larger datasets.
- Introduced SCAN_CONCURRENCY constant to limit the number of concurrent file reads during scans, optimizing performance.
- Refactored file scanning logic to collect candidate files more efficiently and process them in parallel.
- Added new SubagentAttribution interface to standardize the structure of attribution results.
- Expanded unit tests to validate the new file reference retrieval logic for tasks, ensuring accurate member identification across teams.
- Replaced logger.info calls with console.log in TeamMemberLogsFinder for performance logging.
- Simplified environment variable handling in TeamProvisioningService by removing unnecessary context settings.
- Updated stopTeam method to accept a signal parameter for improved process termination control.
- Enhanced model handling in team dialogs to support new context management logic, ensuring consistent behavior across components.
- Adjusted teamSlice to reflect changes in model parsing and context handling, improving overall state management.
- Replaced references to extended context with limit context in TeamProvisioningService and related components, ensuring consistent terminology.
- Updated LaunchTeamDialog and CreateTeamDialog to reflect the new limit context functionality, enhancing user clarity.
- Removed the deprecated ExtendedContextCheckbox component and integrated its functionality into the new LimitContextCheckbox.
- Adjusted teamSlice to accommodate changes in launch parameters, streamlining the context management process.
- Introduced a new system for collecting and prioritizing detection signals for team members, improving accuracy in identifying the correct member from logs.
- Added a `DetectionSignal` interface and a `SIGNAL_PRECEDENCE` array to define the order of reliability for different signal sources.
- Refactored the `attributeSubagent` method to collect signals instead of immediately determining the detected member, allowing for better decision-making based on signal precedence.
- Implemented a new static method `selectBestSignal` to determine the most reliable member signal based on the defined precedence.
- Expanded unit tests to validate the new signal handling logic and ensure correct member identification under various scenarios.
- Updated the detection priority logic for teammate identification to ensure that the teammate ID is recognized as a message sender rather than the file owner.
- Changed the priority threshold from 3 to 2 to allow for more accurate routing of messages based on sender identification.
- Added runtime control API to manage team lifecycle, including launch and stop functionalities.
- Integrated runtime state retrieval to provide current status of teams.
- Enhanced tests to validate the new runtime control features, ensuring proper functionality and error handling.
- Updated type definitions to include runtime API methods for better type safety and clarity in team management.
- Introduced a new caching mechanism for task changes in ChangeExtractorService to improve performance and reduce redundant data fetching.
- Updated TeamInboxWriter to preserve message identity fields such as messageId and timestamp for better deduplication across live and persisted messages.
- Enhanced SendMessageRequest interface to include additional fields like color, toolSummary, and toolCalls for richer message context.
- Improved log file reference resolution in TeamMemberLogsFinder to include filePath for better tracking of log sources.
- Added leadName and leadColor properties to CrossTeamTarget and related interfaces for better team representation.
- Implemented caching for file mentions in TeamMemberLogsFinder to improve performance and reduce redundant file checks.
- Introduced deriveSinceMs method to calculate log search boundaries based on task creation times.
- Updated UI components to display lead information where applicable, enhancing user experience in messaging and logs.
- Introduced a new function to determine the current review state based on task history events, improving the accuracy of review status tracking.
- Updated the requestReview, approveReview, and requestChanges functions to append corresponding review events to the task history, ensuring comprehensive tracking of review actions.
- Refactored task management logic to utilize the new historyEvents structure, replacing the previous statusHistory implementation for better clarity and maintainability.
- Enhanced tests to validate the new review event handling and ensure correct behavior across various task states.
- Removed legacy overlay review state handling from taskStore, simplifying task normalization processes.
- Updated task retrieval methods to directly use normalized task data without fallback to legacy kanban states.
- Eliminated outdated tests related to legacy kanban overlay, focusing on modern task management mechanisms.
- Refactored TeamDataService and TaskBoundaryParser to enhance clarity and maintainability by removing unnecessary complexity.
- Updated related types and interfaces to reflect the removal of legacy support.
- Added maintenance module to the agent-teams-controller, enabling the reconciliation of stale kanban entries and linked inbox comments.
- Implemented garbage collection logic in kanbanStore to remove invalid task references and clean up column orders.
- Updated controller to expose maintenance functionalities, allowing for better task and comment management.
- Enhanced tests to validate the new reconciliation process, ensuring idempotency and correctness in handling stale data.
- Integrated new CLI argument options for worktree and custom arguments in team launch and creation dialogs, improving user flexibility.
- Added new workspace commands for type checking, building, and testing across multiple packages.
- Updated CI workflow to include paths for new packages and utilize workspace commands.
- Refactored MCP server to integrate with the agent-teams-controller, enhancing task management capabilities.
- Improved task boundary detection and logging for MCP tools, ensuring better tracking of task states.
- Updated documentation and prompts to reflect new MCP tool usage, replacing previous teamctl.js references.
- Updated fileMentionsTaskId method to include an 'assumeTeam' parameter for better team detection in logs.
- Added logic to extract team information from process entries and message content, improving task attribution accuracy.
- Enhanced handling of task updates without explicit team names, ensuring correct task processing in solo scenarios.
- Improved overall robustness of log parsing and team identification, facilitating better task management in team environments.
- Implemented deterministic interval scoping in ChangeExtractorService to improve task change retrieval when no specific scope is found.
- Added a new method, extractIntervalScopedChanges, to handle changes based on provided intervals, enhancing the accuracy of change tracking.
- Updated the logic for deriving intervals from task status history, ensuring better handling of task transitions.
- Refactored related code for improved clarity and maintainability, including better type definitions and error handling.
- Updated the `getTaskChanges` method to accept an optional `options` parameter, allowing for more granular control over the retrieval of task changes.
- Introduced new fields in the options object, including `owner`, `status`, `since`, and `intervals`, to improve filtering capabilities.
- Refactored related services and IPC methods to accommodate the new options, enhancing the overall task change management experience.
- Improved type definitions for better clarity and maintainability.
- Added new configuration options for task status change notifications, including toggles for enabling notifications, solo mode restrictions, and customizable target statuses.
- Updated validation logic to accommodate new notification settings, ensuring proper handling of boolean and array types.
- Enhanced UI components to allow users to manage status change notification preferences effectively.
- Implemented backend logic to detect and trigger notifications based on task status changes, improving user awareness of task progress.
- Added a new "Solo mode" feature in the README, allowing a single agent to manage tasks independently, improving token efficiency.
- Updated TeamMemberLogsFinder to include team name in task ID checks, preventing cross-team task collisions.
- Enhanced TeamProvisioningService with clearer task status protocols and execution discipline guidelines to improve task management practices.
- Introduced new callbacks in ChangeReviewDialog for accepting and rejecting new files, streamlining the review process.
- Updated changeReviewSlice to handle file rejection logic, ensuring accurate state management during reviews.
Made-with: Cursor
- Updated WorktreeGrouper to normalize project paths before resolving identities and branches, ensuring consistent behavior across platforms.
- Enhanced TeamDataService to improve lead name resolution and prevent self-notification for comments on lead-owned tasks in solo teams.
- Added new utility functions in TeamDataService for better lead name handling and solo team validation.
- Improved TeamMemberLogsFinder to exclude owner sessions when the owner is the team lead, refining log retrieval logic.
- Updated UI components to reflect changes in team member suggestions based on solo team status.
Made-with: Cursor
- Updated notification configuration to separate settings for lead and user inbox messages, improving user control over notifications.
- Enhanced the handling of inbox message notifications to respect individual inbox settings.
- Introduced new IPC methods for managing watched directories, allowing for more granular file monitoring.
- Improved task management by implementing work intervals for task status transitions, ensuring accurate tracking of task progress.
- Added validation for task creation and configuration, ensuring proper input handling for project paths and task properties.
