Managing Sprints

Complete sprint lifecycle management for AI-agent-driven development.

Usage

/managing-sprints <subcommand> [arguments] [options]

Subcommands

| Subcommand | Description | |------------|-------------| | plan | Create sprint(s) from epic | | execute | Run sprint with parallel agents | | status | Check progress and blockers | | review | Validate completed work |

Subcommand Details

plan

Create sprint project(s) from an epic, grouping features by phase.

/managing-sprints plan <epic-id> [options]

Options:

• --max-sprints N: Limit sprints created (default: no limit)
• --dry-run: Preview without creating

Details: planning.md

execute

Launch parallel AI agents to implement sprint work.

/managing-sprints execute <sprint-id> [options]

Options:

• --dry-run: Preview execution plan only
• --max-agents N: Limit parallel agents (default: 4)
• --phase: Execute specific phase only (foundation|features|integration)

Details: executing.md

status

Check sprint progress, active agents, and blockers.

/managing-sprints status [sprint-id] [options]

Options:

• --active: Show only active sprints
• --detailed: Include agent activity
• --team: Filter by team

Details: monitoring.md

review

Validate completed work against acceptance criteria.

/managing-sprints review <sprint-id>

Details: reviewing.md

Execution Philosophy

Three-Phase Model

1. Foundation Phase (0-10% of work)

   • Database migrations
   • Core infrastructure
   • Shared utilities
   • Sequential execution - must complete before Features

2. Features Phase (70-85% of work)

   • Main functional implementation
   • UI components
   • API endpoints
   • Maximum parallelization - run as many agents as safe

3. Integration Phase (10-20% of work)

   • E2E testing
   • Documentation
   • Performance tuning
   • After Features complete

Parallelization Strategy

Safe to parallelize:

• Different file areas (frontend vs backend)
• Independent features
• Non-overlapping components

Must serialize:

• Same file modifications
• Dependent features
• Shared state changes

Agent Assignment

Each agent receives:

• Coherent set of related issues
• Complete functional domain
• Balanced complexity

Guidelines:

• Max 4 parallel agents (avoid resource contention)
• Group by stack area (frontend, backend, etc.)
• Group by codebase area (same files → same agent)
• Keep tightly-coupled issues together

PM Tool Integration

Uses pm-context for all operations:

• Fetch sprint/project details
• Query work items
• Update status
• Add progress comments

Example Workflow

# 1. Plan sprints from epic
/managing-sprints plan CCC-123 --max-sprints 3

# 2. Execute first sprint
/managing-sprints execute CCC-123.S01

# 3. Monitor progress
/managing-sprints status CCC-123.S01 --detailed

# 4. Review when complete
/managing-sprints review CCC-123.S01

# 5. Execute next sprint
/managing-sprints execute CCC-123.S02

Output Examples

Plan Output

📋 Sprint Planning: User Authentication Epic

🔍 Analyzing 12 features, 43 tasks...

📊 Sprint Distribution:
Sprint 1 (CCC-123.S01):
  - Foundation: JWT infrastructure, user schema
  - Features: Login, signup (parallel)
  - 8 features, 24 tasks

Sprint 2 (CCC-123.S02):
  - Features: OAuth, password reset (parallel)
  - 3 features, 12 tasks

Sprint 3 (CCC-123.S03):
  - Integration: E2E tests, documentation
  - 1 feature, 7 tasks

✅ Created 3 sprint projects

Execute Output

🚀 Sprint Execution: CCC-123.S01

Phase 1: Foundation
⏳ Agent-1: Database schema migration
✅ Complete (15 min)

Phase 2: Features (4 parallel agents)
⏳ Agent-1: Login form & API
⏳ Agent-2: Signup flow
⏳ Agent-3: Session management
⏳ Agent-4: Email verification

[Progress updates as agents work...]

Phase 3: Integration
⏳ Agent-1: E2E auth tests
✅ Complete

📊 Sprint Complete!
- 8 features implemented
- 24 tasks completed
- 0 blockers

Error Handling

| Error | Recovery | |-------|----------| | File conflict detected | Reassign to same agent | | Agent failure | Retry up to 2 times | | Dependency not met | Block and notify | | PM tool error | Retry with backoff |

Invocation Control

This skill has disable-model-invocation: true because:

• Execution launches multiple agents with side effects
• User should explicitly control when sprints run
• Prevents accidental execution from context

User must explicitly invoke: /managing-sprints execute ...