process-tasklist

Prompt file imported from lizthrilla/jetiller (.cursor/commands/process-tasklist.md). Copyright stays with the author.

Rule: Processing a Task List with User Oversight

Goal

To guide an AI assistant in implementing a task list step-by-step, with the user overseeing and approving progress at each stage. The AI handles the implementation while the user maintains control over the direction and quality of the work.

Process

1. Receive Task List: The user provides a link to an existing task list (e.g., /.tasks/tasklist-[feature-name].md).
2. Review and Confirm: The AI reads the task list and presents a summary to the user, confirming:
   • Total number of tasks and subtasks
   • Any dependencies or prerequisites
   • Suggested order of execution
3. Work Through Tasks: For each task:
   • Announce: Tell the user which task/subtask is being worked on.
   • Implement: Complete the subtask (create/modify files, write code).
   • Report: Summarize what was done and show key changes.
   • Checkpoint: Ask the user to confirm before moving to the next subtask.
4. Handle Issues: If a blocker or question arises, pause and ask the user for guidance before proceeding.
5. Mark Progress: Update the task list file to mark completed items as the work progresses.
6. Final Review: Once all tasks are complete, provide a summary of all changes made.

Checkpoint Questions

At each checkpoint, the AI should ask one of the following (adapt based on context):

• "Subtask 1.1 is complete. Ready to proceed to Subtask 1.2?"
• "I've finished Task 1. Would you like to review before I start Task 2?"
• "I ran into an issue with [X]. How would you like me to proceed?"
  • a) Try an alternative approach
  • b) Skip this subtask for now
  • c) Let me provide more context
• "The implementation differs slightly from the spec because [reason]. Is this acceptable?"
  • a) Yes, proceed
  • b) No, let's adjust

Progress Tracking Format

As tasks are completed, update the task list with status indicators:

### Task 1: Create the data model for user preferences ✅

**Status:** Complete

#### Subtask 1.1: Define the UserPreferences interface ✅
#### Subtask 1.2: Create default preference values ✅

---

### Task 2: Build the UserPreferences component 🔄

**Status:** In Progress

#### Subtask 2.1: Create the component file and basic structure ✅
#### Subtask 2.2: Implement the preference editing UI ⏳ (current)

Status Icons:

• ✅ Complete
• 🔄 In Progress
• ⏳ Current subtask
• ⏸️ Paused/Blocked
• ❌ Skipped

Implementation Guidelines

When implementing each subtask, the AI should:

1. Read First: Review any existing files that will be modified.
2. Follow Patterns: Match the existing code style and conventions in the project.
3. Minimal Changes: Only make changes required by the subtask — avoid scope creep.
4. Show Work: Display the key code changes to the user (not every line, but the important parts).
5. Verify: Run any relevant linters or type checks after making changes.

Handling Blockers

If the AI encounters an issue:

1. Stop: Do not attempt to work around the issue without user input.
2. Explain: Clearly describe what the blocker is.
3. Offer Options: Provide 2-3 possible ways to proceed (lettered for easy selection).
4. Wait: Do not continue until the user responds.

Example:

🚧 **Blocker Encountered**

While working on Subtask 2.1, I noticed that the `UserContext` provider referenced in the PRD doesn't exist yet.

How would you like to proceed?
- a) Create the UserContext provider now (adds a new subtask)
- b) Skip the context integration for now and add a TODO comment
- c) Point me to an existing context I should use instead

Summary Report Format

At the end of the session (or when requested), provide a summary:

## Session Summary

**Task List:** `/.tasks/tasklist-user-preferences.md`

### Completed
- ✅ Task 1: Create the data model (2 subtasks)
- ✅ Task 2: Build the UserPreferences component (3 subtasks)

### In Progress
- 🔄 Task 3: Add preferences to settings page (1/2 subtasks done)

### Not Started
- Task 4: Write unit tests
- Task 5: Update documentation

### Files Modified
- `src/types/UserPreferences.ts` (created)
- `src/constants/defaults.ts` (modified)
- `src/components/UserPreferences.tsx` (created)
- `src/components/UserPreferences.module.css` (created)

### Notes
- Subtask 3.2 is blocked pending design review
- Suggested adding error boundary in Task 2 (approved by user)

User Control Commands

The user can direct the workflow at any time with commands like:

• "Skip this" — Skip the current subtask and move on.
• "Pause" — Stop after the current subtask and wait.
• "Show me the task list" — Display current progress.
• "Go back to [task/subtask]" — Revisit a previous item.
• "What's next?" — Show upcoming tasks/subtasks.
• "Summarize progress" — Get the summary report.

Target Audience

The user overseeing the work may have varying levels of technical expertise. Therefore:

• Be Clear: Explain what you're doing in plain language.
• Be Concise: Don't over-explain simple changes.
• Be Responsive: Adapt to user feedback and preferences.
• Be Transparent: If something is unclear or risky, say so.

The goal is a collaborative workflow where the AI does the heavy lifting but the user stays informed and in control.