Refining Work Items

Transform rough ideas or existing work items into well-structured, AI-agent-ready specifications using proven templates.

Usage

/refining-work-items [item-id] [options]
/refining-work-items --new --type epic --team "TeamName"

Arguments

Item ID (optional):

• Existing work item to refine (e.g., CCC-123, PROJ-456, #42)
• Or full URL from PM tool
• If omitted with --team, creates new item

Options:

• --type or type: epic | epic-lite | feature | task | bug | chore
• --analyze-codebase or analyze codebase: Add codebase analysis for technical context
• --interactive or interactive: Step through sections with guided prompts
• --validate-only or validate only: Check readiness without making changes
• --team followed by name: Target team (required for create mode)
• --parent or epic followed by ID: Link to parent item
• --create-subtasks: Auto-generate child items (features/tasks)

Examples

# Refine existing epic with full PRD template
/refining-work-items CCC-123 --type epic

# Quick 1-pager epic refinement
/refining-work-items CCC-123 --type epic-lite

# Refine with codebase analysis
/refining-work-items CCC-123 --type epic --analyze-codebase

# Create new feature for a team
/refining-work-items --new --type feature --team Chronicle

# Refine feature and auto-create subtasks
/refining-work-items PROJ-456 --type feature --create-subtasks

# Interactive bug report refinement
/refining-work-items --new --type bug --team Chronicle --interactive

Workflow

Step 1: Setup

1. Detect PM Tool: Load pm-context to detect project's PM tool
2. Parse Arguments: Extract item ID, type, and options from input
3. Determine Mode:
   • Item ID provided → Refine Mode
   • No ID + team provided → Create Mode
   • Neither → Error with guidance

Step 2: Context Gathering

For Refine Mode:

• Fetch item via pm-context get_item(id)
• Extract current title, description, status
• Check parent/child relationships
• Identify existing labels and type

For Create Mode:

• Prompt for initial idea/description
• Auto-generate title suggestion
• Determine type from keywords or prompt

Step 3: Type Detection & Template Selection

If type not specified, detect from:

1. Explicit labels (epic, feature, bug, etc.)
2. Parent relationships (child of epic → feature)
3. Keywords in title/description
4. Interactive prompt if ambiguous

Load appropriate template:

Type	Template
epic	epic-template.md
epic-lite	epic-lite.md
feature	feature-template.md
task/bug/chore	task-template.md

Step 4: Readiness Assessment

Check item against template requirements:

Epic Readiness:

• Problem statement defined
• User stories or requirements
• Acceptance criteria (measurable)
• Technical requirements identified
• Success metrics defined

Feature Readiness:

• Clear, specific title
• User story or functional requirement
• 3+ acceptance criteria
• Complexity estimated
• Parent epic linked (if applicable)

Task/Bug Readiness:

• Clear title
• Context/description
• Definition of done
• Priority set

Report gaps and offer to fill interactively.

Step 5: Template Population

Interactive Mode (--interactive): Walk through each template section with prompts.

Auto Mode (default):

• Extract information from existing content
• Infer missing sections where possible
• Flag sections requiring input

Step 6: Codebase Analysis (Optional)

When --analyze-codebase is specified:

1. Extract technical areas from description
2. Launch parallel analysis agents:

   For each technical area:
   - Search for existing implementations (Glob)
   - Find patterns and constraints (Grep)
   - Check ai_docs/knowledge/ for context
   - Return brief technical context (1 paragraph)
   

3. Synthesize findings into Technical Context section

Step 7: Output

Update existing item:

pm-context.update_item(id, {
    title: refined_title,
    description: populated_template,
    labels: [type_label, "refined"]
})

Create new item:

pm-context.create_item({
    type: item_type,
    title: title,
    description: populated_template,
    parent: parent_id,
    labels: [type_label]
})

Step 8: Validation

• All required template sections populated
• Acceptance criteria are measurable
• Technical context sufficient for AI agents
• Dependencies identified
• Item linked to parent (if applicable)

Output Examples

Epic Refinement

🔍 Analyzing work item CCC-123...
✅ Found: "Add user authentication system"
📋 Type: Epic (detected from labels)

📝 Gathering Context:
  ✅ Extracted requirements from description
  ✅ Found 3 related items
  ✅ Identified tech stack: React + FastAPI

🤖 AI-Optimization:
  📦 Estimated 4-6 features across 3 phases
  🔧 Foundation: JWT infrastructure, user schema
  ⚡ Features: Login, signup, OAuth (parallel)
  🔗 Integration: Security testing, docs

✅ Epic refined: CCC-123
🔗 View: [URL]
📊 Ready for /breaking-down-work

Feature Creation

🆕 Creating New Feature

📝 What feature do you want to build?
> User can reset their password via email

🏷️ Suggested title: "Password Reset via Email"
🔗 Parent epic: CCC-123 (User Authentication)

📋 Generating feature specification...
  ✅ User story defined
  ✅ 5 acceptance criteria
  ✅ Technical approach outlined
  ✅ Subtasks identified (3)

✅ Feature created: CCC-456
🔗 View: [URL]

Next Steps

After refining:

• Epic: Run /breaking-down-work <id> to create features/tasks
• Feature: Run with --create-subtasks or manually break down
• Task/Bug: Ready for sprint assignment

Error Handling

Error	Solution
"Item not found"	Verify ID format and access permissions
"Insufficient context"	Use --interactive mode for guided input
"Type ambiguous"	Specify --type explicitly
"Parent not found"	Verify parent ID exists