tasks-load（任务负载）

Load Project Tasks

Load tasks from projects/$ARGUMENTS/tasks/ into the current Claude Code session.

Tasks are stored in session subdirectories to preserve history across /clear commands:

projects/$ARGUMENTS/tasks/
├── {session-1-uuid}/
│   ├── 1.json
│   └── 2.json
└── {session-2-uuid}/
    ├── 1.json
    └── 2.json

Process

Step 1: Validate Project

Check if the project exists and has task files:

find projects/$ARGUMENTS/tasks -name "*.json" 2>/dev/null | head -5

If no task files exist, report: "No tasks found in projects/$ARGUMENTS/tasks/"

List available sessions:

ls -dt projects/$ARGUMENTS/tasks/*/ 2>/dev/null | head -10

Step 2: Set Active Project

Create the active project marker:

echo "$ARGUMENTS" > .claude-active-project

This ensures any new tasks created will sync back to this project.

Step 3: Load Tasks

Find and load all task JSON files from ALL session directories:

find projects/$ARGUMENTS/tasks -name "*.json" -type f

For each JSON file found:

1. Read the task JSON file
2. Use TaskCreate to recreate the task with:

- subject from JSON - description from JSON - activeForm from JSON - metadata: {"project": "$ARGUMENTS"}

1. If the task was already completed (status: "completed"), use TaskUpdate to mark it completed

Step 4: Report

After loading all tasks, report:

Loaded X tasks from projects/$ARGUMENTS/tasks/
- Sessions found: N
- Pending: Y
- Completed: Z

Active project set to: $ARGUMENTS
New tasks will automatically sync to this project.

Notes

• Tasks are recreated with new IDs in the current session
• The original task IDs from the project are not preserved
• Tasks from ALL sessions are loaded (full history)
• Task dependencies (blocks/blockedBy) are NOT currently preserved across load/sync cycles
• Use TaskList to see the loaded tasks