Tool Specification: mshtools-todo_write
Overview
Structured task list management system for tracking progress, organizing complex tasks, and maintaining state across multi-step operations. Creates hierarchical todo lists with status tracking.
JSON Schema
{
"type": "object",
"properties": {
"todos": {
"type": "array",
"items": {
"type": "object",
"properties": {
"id": {
"type": "string",
"description": "Unique identifier for the task"
},
"content": {
"type": "string",
"description": "Task description"
},
"status": {
"type": "string",
"enum": ["pending", "in_progress", "completed"],
"description": "Current status of the task"
},
"priority": {
"type": "string",
"enum": ["high", "medium", "low"],
"description": "Task priority level"
}
},
"required": ["id", "content", "status", "priority"]
}
}
},
"required": ["todos"]
}
Streaming Mechanism
- Transport: Direct state update to session storage
- Persistence: Stored in
/mnt/kimi/.store/.todo.jsonl(634 bytes typical) - Update Model: Full replacement of todo list (not incremental)
- Access: Read via
todo_read(no parameters required)
Integration Architecture
Storage Layer
- Path:
/mnt/kimi/.store/.todo.jsonl - Format: JSON Lines (append-only log)
- Scope: Session-scoped (persists across turns but not conversations)
- Size: ~634 bytes typical
State Management
# Internal representation
{
"todos": [
{
"id": "1",
"content": "Initialize project",
"status": "completed",
"priority": "high"
},
{
"id": "2",
"content": "Build components",
"status": "in_progress",
"priority": "high"
}
]
}
Operational Rules
When to Use
- Complex multi-step tasks (3+ distinct actions)
- Non-trivial tasks requiring planning
- User explicitly requests todo list
- Multiple tasks (numbered or comma-separated)
- After receiving new instructions
When NOT to Use
- Single straightforward task
- Trivial tasks (<3 steps)
- Purely conversational tasks
Task States
- pending: Not started
- in_progress: Currently working (only ONE at a time allowed)
- completed: Finished successfully
Management Rules
- Update status live while working
- Complete tasks immediately after finishing
- Don't batch completions
- Remove irrelevant tasks
- Mark as completed only when: fully accomplished, no errors, final implementation, dependencies found
Usage Patterns
Creating Initial List
{
"todos": [
{"id": "1", "content": "Read skill documentation", "status": "completed", "priority": "high"},
{"id": "2", "content": "Create project structure", "status": "in_progress", "priority": "high"},
{"id": "3", "content": "Implement features", "status": "pending", "priority": "medium"}
]
}
Updating Progress
{
"todos": [
{"id": "1", "content": "Read skill documentation", "status": "completed", "priority": "high"},
{"id": "2", "content": "Create project structure", "status": "completed", "priority": "high"},
{"id": "3", "content": "Implement features", "status": "in_progress", "priority": "medium"}
]
}
System Integration
- OK Computer Only: Not available in Base Chat
- Proactive Use: System encourages frequent checking via
todo_read - Workflow Integration: Used at conversation start, before new tasks, after completions