GTD Next Action

Reference: See things3 skill for task requirements, tag taxonomy, and MCP operations.

Identify and execute/assist with the highest priority next action from Today.

Auto-Execute Scope: Research tasks, email drafts, and code changes.

Workflow

1. Get Today's Tasks

mcp__things__get_today

2. Check for Daily Review

If any task belongs to Project: ☕️ Daily Review:

• The Daily Review project is active
• Work through it using the gtd-daily-review skill workflow
• Process project tasks in order, not individual prioritization

3. Prioritize Tasks

For non-Daily Review tasks, sort by (highest priority first):

1. Agent/Queued - Tasks explicitly queued for agent work
2. Calendar-related - Tasks tied to today's meetings
3. Quick wins (2m, 5m) - Build momentum
4. Time-blocked - Scheduled for specific times
5. Deep work (25m, 1h+) - Focused tasks

4. Select and Analyze Top Action

Is it tagged "Agent/Queued"?
├── YES → Claim it and execute
└── NO → Continue analysis...

Is it part of a project with other tasks?
├── YES → Read project context, understand scope
└── NO → Treat as standalone task

Can Claude help directly?
├── Research → Execute: web search, summarize
├── Email → Execute: draft, present for review
├── Code → Execute: implement, test, report
├── Writing → Execute: draft content
└── Physical/human → Assist: provide context

Is it blocked?
├── YES → Tag "Agent/Blocked", pick next task
└── NO → Proceed

Needs human verification when done?
├── YES (code, important decisions) → Use Agent/Needs Review
└── NO (research, simple tasks) → Complete directly

5. Claim Agent Tasks

For tasks tagged Agent/Queued, claim before working:

mcp__things__update_todo(id, tags=["Agent", "Working"])

6. Execute or Assist

For AI-automatable tasks: Just do it - research, draft, implement.

For human-required tasks: Provide context, guidance, resources.

7. Track Progress

As you complete checklist items, report progress to the user:

Working on "Research competitor pricing":
- [x] Identify top 5 competitors (done)
- [x] Document pricing tiers (done)
- [ ] Create comparison spreadsheet (in progress)

Please check off completed items in Things.

Update task notes with findings and decisions (not checkboxes):

mcp__things__update_todo(id, notes="Context: [original context]

---
PROGRESS:
- Found 5 competitors: A, B, C, D, E
- Pricing ranges from $10-50/mo")

8. Handle Completion

Blocked:

mcp__things__update_todo(id, tags=["Agent", "Blocked"], notes="Blocked: [reason]")

Needs Review:

mcp__things__update_todo(id, tags=["Agent", "Needs Review"], notes="Done: [summary]")

Complete:

mcp__things__update_todo(id, completed=true, tags=[])

9. Report and Continue

Completed: "Research competitor pricing" [5m]

Results:
- [Summary of what was done]
- [Key findings]

Next up: "Draft proposal outline" [25m]
Continue? [Y/n]

MCP Tools Reference

Guidelines

• Check for Daily Review first - If active, work through that project
• Claim before working - Always claim Agent/Queued tasks first
• Execute directly - For research, drafts, code - just do it
• Track progress - Report checklist completion to user, update notes with findings
• Use Blocked when stuck - Don't spin; mark blocked and explain
• Use Needs Review for consequential work - Code, important decisions
• Continue momentum - After completing, offer next task