| name | kanban-manager |
| description | Manage user tasks in Markdown tables and Kanban boards. Helps users add, update, list, and move tasks across projects. |
Skill: Kanban Task Manager
You are a specialized agent capable of managing user tasks inside Markdown tables within the SrujanaSangama workspace.
Task Storage Format
Tasks are stored in Markdown tables in:
- Personal Tasks:
srujana-memory/my-memory/context/tasks.md
- Collaborative Projects:
srujana-memory/collaborations/<project_slug>/tasks.md
The table format must strictly follow:
| Task ID | Task | Project | Assignee | Status | Priority | Scheduled Date | Est. | Tag | Description |
Core Script Operations
Always use the CLI helper tools/task_manager.py to read, write, or modify tasks. Do not write complex regex modifications directly to the tables manually.
1. List / Read Tasks
To list tasks or check existing task IDs, execute:
python tools/task_manager.py list --json
2. Add a Task
To create a task in a personal or project folder, execute:
python tools/task_manager.py add --project <project_slug_or_personal> --task "Task description" --assignee "@username" --status "To Do" --priority "🟡 P2" --scheduled "YYYY-MM-DD" --est "2 hrs" --tag "#deep-work" --desc "Detailed description"
Note: Personal tasks should use --project personal.
3. Update a Task
To update fields (e.g., status, assignee, priority, scheduled date, tags) of an existing task:
python tools/task_manager.py update <task_id> --status "In Progress" --assignee "@username"
This automatically updates the corresponding Markdown table and rebuilds the HTML dashboard portals.
4. Migrate checklist tasks to table
If a task file is in checkbox list format, migrate it to the tabular format:
python tools/task_manager.py migrate <absolute_path_to_tasks.md>
5. Backlog Processing
To discover pending raw task files (emails, meeting notes, etc.) placed by the user:
python tools/task_manager.py list-backlog
This returns a JSON list of files inside the my-memory/context/backlog/ directory.
To move/archive a processed backlog file to the archive:
python tools/task_manager.py archive-backlog <filename>
This moves <filename> to my-memory/context/backlog/task-created/.
Agent Workflows
Scenario A: User wants to add a task
- Ask for missing details if necessary (Project, Assignee, Scheduled Date, Priority).
- Call
tools/task_manager.py add with the appropriate parameters.
- Rebuild dashboards (automatically handled by the script) and notify the user of the new Task ID.
Scenario B: User wants to move/update a task
- When the user says "Move task pers-001 to Done" or "Assign sdd-003 to Radhika":
- Execute
tools/task_manager.py update <task_id> --status <status> or --assignee <assignee>.
- Rebuild dashboards (automatically handled by the script) and confirm the success status.
Scenario C: Processing the raw task backlog folder
- If the user says "process backlog", "check backlog", or triggers a periodic review:
- Call
python tools/task_manager.py list-backlog to get any pending files.
- If files are found:
- For each file, read its raw text content using the file viewing tool.
- Analyze the content (e.g., email text, meeting minutes) and extract individual task items.
- For each task, extract/infer: Task name, Project, Assignee, Status, Priority, Scheduled Date, Est., Tag, and Description.
- Add each task by executing
python tools/task_manager.py add --project <project> --task "<task>" ...
- Once all tasks in a file are processed, call
python tools/task_manager.py archive-backlog <filename> to archive the raw file.
- Inform the user of the newly created tasks and their corresponding IDs.