Capture discussion/proposal comments as separate task type instead of ignoring them

[biwakonbu]

Problem

Review comments that are discussions, proposals, or questions (not code fixes) are currently ignored by the task generation system. These important conversations get lost and are never tracked.

Current Behavior

Comments that get ignored:

• "What do you think about using X pattern instead?"
• "Should we consider performance implications here?"
• "Have you thought about edge case Y?"
• "I suggest we refactor this differently - thoughts?"
• Discussion threads about design decisions
• Proposal comments without direct code changes

Result:

• Important discussions are not tracked
• Developer may forget to respond
• Reviewers feel their questions are ignored
• No clear workflow for handling non-implementation feedback

Desired Behavior

Task Type Classification

Create a new task type for discussion/proposal comments:

Implementation Tasks (existing):

• Require code changes
• Can be verified programmatically
• Have clear done criteria
• Example: "Fix null pointer bug in handler"

Discussion Tasks (new):

• Require human response/decision
• Cannot be auto-verified
• Need reviewer acknowledgment
• Example: "Respond to performance optimization proposal"

Workflow for Discussion Tasks

# System detects discussion comment and creates special task
$ reviewtask
✓ Generated 3 implementation tasks
✓ Generated 2 discussion tasks

📋 Discussion tasks require your response on GitHub

# Show discussion task
$ reviewtask show abc123
Task: Respond to reviewer's refactoring proposal
Type: discussion
Priority: medium
Status: todo

💬 Reviewer Question:
"Have you considered using the strategy pattern here? 
It might make this more extensible."

📋 Next steps:
  • Read the full discussion: [GitHub Thread URL]
  • Reply to the comment on GitHub with your thoughts
  • Run 'reviewtask update abc123 done' after responding

# User responds on GitHub, then marks as done
$ reviewtask update abc123 done
✓ Updated task 'abc123' to 'done'

📋 Note: Discussion tasks are resolved when marked done

Implementation Strategy

Phase 1: Detect Discussion Comments

• AI prompt enhancement to classify comment types
• Distinguish between "implement this" vs "what about this?"
• Identify questions, proposals, and open discussions
• Mark comments requiring human dialogue

Phase 2: Create Discussion Task Type

• Add type field to Task struct: "implementation" | "discussion"
• Store discussion tasks separately or with type flag
• Display discussion tasks distinctly in reviewtask show
• Filter tasks by type in reviewtask status

Phase 3: Guide User Response

• Show discussion tasks with "respond on GitHub" guidance
• Include direct link to comment thread
• Suggest marking done after responding
• Track response status

Phase 4: Future Enhancement (Auto-resolution)

• Detect when user has replied on GitHub
• Auto-mark discussion tasks as done when response posted
• Auto-resolve thread after user response (optional)
• Notify if discussion continues (new replies)

Example AI Task Generation

Current (Ignored)

{
  "comment": "What do you think about caching this result?",
  "action": "IGNORED - no clear implementation task"
}

Proposed (Captured)

{
  "id": "task-disc-1",
  "type": "discussion",
  "category": "proposal",
  "title": "Respond to caching suggestion",
  "description": "Reviewer proposes adding caching. Discuss feasibility and trade-offs.",
  "reviewer_question": "What do you think about caching this result?",
  "requires_response": true,
  "implementation_task": false,
  "source_comment_id": 12345,
  "url": "https://github.com/..."
}

Task Display Format

Implementation Task (Existing)

📝 Fix validation error handling
Type: implementation
Priority: high
Status: doing

□ Add null check before validation
□ Return proper error message

Discussion Task (New)

💬 Respond to performance optimization proposal
Type: discussion
Priority: medium  
Status: todo

Reviewer asks: "Should we batch these database calls?"

Action needed:
→ Respond on GitHub: [link to thread]
→ Mark done after responding

Benefits

For Developers

• ✅ Never miss important discussions
• ✅ Clear action item to respond
• ✅ Track discussion resolution like other tasks
• ✅ Guided workflow for non-code feedback

For Reviewers

• ✅ Confidence that questions won't be ignored
• ✅ See when developer has responded (future: auto-detect)
• ✅ Clear signal that feedback was acknowledged
• ✅ Better collaboration on design decisions

For Teams

• ✅ Track both implementation and discussion progress
• ✅ Ensure all review feedback is addressed
• ✅ Prevent lost conversations
• ✅ Complete PR review accountability

Configuration

Add settings for discussion task handling:

{
  "ai_settings": {
    "process_discussion_comments": true,
    "discussion_task_priority": "medium",
    "auto_resolve_discussions": false
  }
}

Example Use Cases

Case 1: Design Question

Reviewer: "Why did you choose approach X over Y?"
→ Discussion Task: "Explain design decision to reviewer"
→ User responds on GitHub with rationale
→ Mark task as done

Case 2: Proposal

Reviewer: "I suggest using dependency injection here"
→ Discussion Task: "Consider dependency injection proposal"  
→ User discusses pros/cons on GitHub
→ May create follow-up implementation task
→ Mark discussion as done

Case 3: Open Question

Reviewer: "How does this handle edge case Z?"
→ Discussion Task: "Answer edge case handling question"
→ User explains on GitHub or creates implementation task
→ Mark discussion as done

Success Criteria

• Zero important discussions are ignored
• Users have clear action items for responding to questions
• Discussion tasks are visibly different from implementation tasks
• Workflow guides user to respond on GitHub
• Can filter/view tasks by type (implementation vs discussion)
• Future: Auto-detect when user has responded

Future Enhancements (Out of Scope for Initial Implementation)

1. Auto-Response Detection: Detect when user has replied on GitHub and auto-mark done
2. Thread Continuation: Create new discussion task if reviewer replies again
3. AI-Assisted Drafts: Generate suggested responses for common questions
4. Discussion Summary: Summarize multi-comment discussions into single task
5. Auto-Resolution: Auto-resolve discussion threads when marked done

Related Issues

• Relates to Support detecting resolved review comments to avoid redundant task generation #185 (resolved comment detection) - helps track which discussions are closed
• Relates to Simplify workflow and add intelligent operation guidance to maintain stability #187 (workflow guidance) - guidance would show how to handle discussions
• Relates to task deduplication - avoid duplicate discussion tasks for same question

[biwakonbu]

This issue proposes adding discussion/proposal task types, which is a valuable enhancement that goes beyond #192 (comprehensive comment analysis).

Relationship to new issues:

• [Phase 2] Implement comprehensive comment analysis with AI impact assessment #192 ensures ALL comments are analyzed (including questions/discussions)
• This issue (Capture discussion/proposal comments as separate task type instead of ignoring them #188) proposes a separate task TYPE to handle discussions differently

Recommendation:
Keep this issue open as a future enhancement. After #192 is implemented, we can add discussion task types as a follow-up feature.

The implementation would add:

• Task type field: 'implementation' vs 'discussion'
• Different workflow for discussion tasks (respond on GitHub)
• Auto-detection when user has replied

This is a separate concern from impact assessment and should be tracked independently.