paulnsorensen/update

Todoist Update

Walk through active tasks with the user to find completed work, irrelevant items, and tasks that need better formatting. This is routine hygiene — the counterpart to /triage (which handles overdue debt).

Flow

Step 1: Choose Scope

Ask the user what to review via AskUserQuestion:

What would you like to review?
(A) All projects — full sweep
(P) Pick a project — I'll list them
(F) Flagged tasks only — P1/P2 tasks (are priorities still accurate?)

If they pick (P), fetch projects via mcp__todoist__find-projects and let them choose.

Step 2: Fetch Tasks

Spawn todoist-fetch to retrieve tasks in a fresh context window:

Agent(subagent_type: "todoist-fetch", prompt: "Fetch active (non-overdue) tasks for [scope]. Exclude overdue tasks. Include task IDs, titles, projects, priorities, due dates, creation dates, descriptions, and labels. Flag: tasks with no due date, P1/P2 tasks, tasks created 90+ days ago.")

Focus on:

• Tasks with no due date (dormant — still relevant?)
• Tasks due in the future (still planned correctly?)
• High-priority tasks (P1/P2 — priority still accurate?)

Step 3: Present for Review

Show 5 tasks at a time:

### Task 1/N: [Task Title]
- Project: [project]
- Priority: [p1-p4]
- Due: [date or "no date"]
- Created: [date] (X days ago)
- Description: [if any]

Actions: (D)one | (X)Delete — irrelevant | (R)eword | (P)riority change [p1-p4] | (S)chedule [when?] | (K)eep | (?)Research

Use AskUserQuestion for decisions. Accept shorthand: d, x, r, p2, s next week, k, ?.

Step 4: Handle Rewording

If the user picks (R), ask what the new title should be — or offer a suggestion based on best practices:

Current: "dentist"
Suggested: "Call dentist to schedule cleaning"
Use suggestion, or type your own:

The todoist-scribe agent will validate the reworded title against formatting rules.

Step 5: Handle Research

If ?, invoke the research skill for context:

Skill(skill="todoist-flow:research", args="Context for: '[task title]'. [description]. Help the user decide if this task is still relevant.")

Re-present the task after research returns.

Step 6: Execute

Batch decisions per group of 5 and run through the write pipeline (distill → scribe → QA):

1. Validate reasoning — spawn todoist-distill:

Agent(subagent_type: "todoist-distill", prompt: "Validate these update decisions against user intent: [decisions with task data and user choices]")

2. Format commands — spawn todoist-scribe with validated plan:

Agent(subagent_type: "todoist-scribe", prompt: "Format these validated update operations as MCP commands: [distill's validated plan]")

3. Verify and execute — spawn todoist-qa:

Agent(subagent_type: "todoist-qa", prompt: "Verify and execute: [scribe's formatted commands]. Original intent: [distill's validated plan]")

Step 7: Running Stats

After each batch show progress:

Progress: [X/Y tasks reviewed]
Done: X | Deleted: X | Reworded: X | Reprioritized: X | Scheduled: X | Kept: X

Ask to continue or stop.

Step 8: Summary

## Update Complete — [Project Name or "All Projects"]

| Action | Count |
|--------|-------|
| Marked done | X |
| Deleted (irrelevant) | X |
| Reworded | X |
| Reprioritized | X |
| Scheduled | X |
| Kept as-is | X |

Tasks reviewed: X of Y

Priority Audit Mode

When scope is (F) flagged tasks, focus the review on priority accuracy:

• P1 tasks: "Is this genuinely urgent and important today? Or has the urgency passed?"
• P2 tasks: "Is this still this week's focused work?"
• Show current P1 count — if >3, flag inflation: "You have X tasks at P1. Ideally 1-3. Which are truly urgent today?"

Key Principles

• Incremental, not bankruptcy — the user explicitly doesn't want to archive everything. Respect that by reviewing task-by-task.
• Completed work matters — the user may have done tasks without marking them complete. This is the most satisfying part of the review.
• Formatting is secondary — if a task is correctly titled and prioritized, don't suggest changes just for polish. Only reword tasks that are genuinely unclear.
• No guilt — deletion is healthy. Don't make the user feel bad about removing irrelevant tasks. Frame it as "this served its purpose" or "priorities shifted."