priority-management
SKILL.md
Priority Management
Purpose
Determine the optimal sequence and priority of tasks based on dependencies, business value, risk, and resource availability to maximize project progress and value delivery.
When to Use
- Sprint planning
- New urgent request arrives
- Resource constraints change
- Scope changes
- Blocker impacts schedule
- Stakeholder priority changes
Prerequisites
- Task breakdown available
- Dependencies mapped
- Business value understood
- Resource availability known
Process
Step 1: Assess Task Attributes
For each task, evaluate:
- Business value (High/Medium/Low)
- Urgency (Critical/High/Normal/Low)
- Dependencies (blocking others?)
- Risk (technical uncertainty)
- Effort (S/M/L/XL)
- Required skills (which agent)
Step 2: Apply Prioritization Matrix
Priority Score = (Value × 3) + (Urgency × 2) + (Blocking × 2) - (Risk × 1)
Where:
- Value: High=3, Medium=2, Low=1
- Urgency: Critical=4, High=3, Normal=2, Low=1
- Blocking: Blocks 3+=3, Blocks 1-2=2, Blocks 0=1
- Risk: High=3, Medium=2, Low=1
Step 3: Consider Dependencies
Dependency rules:
1. Blocked tasks cannot be prioritized until blocker resolves
2. Tasks blocking many others get priority boost
3. External dependencies flagged for early start
4. Critical path tasks get highest attention
Step 4: Balance Agent Workload
Distribution rules:
1. No agent > 3 concurrent tasks
2. Mix task sizes (not all XL)
3. Consider skill match
4. Allow slack for interrupts
Step 5: Create Priority Queue
Output priority-sorted task list:
1. P0 - Critical: Do immediately
2. P1 - High: Do this sprint
3. P2 - Normal: Do when P0/P1 clear
4. P3 - Low: Backlog
Inputs
| Input | Type | Required | Description |
|---|---|---|---|
| task_breakdown.json | JSON | Yes | All tasks |
| dependency_graph.json | JSON | Yes | Task dependencies |
| agent_availability | JSON | Yes | Agent capacity |
| business_priorities | JSON | Optional | Stakeholder input |
Outputs
| Output | Type | Description |
|---|---|---|
| priority_queue.json | JSON | Sorted task list |
| sprint_backlog.md | Markdown | Sprint plan |
| agent_assignments.json | JSON | Task distribution |
StudyAbroad-Specific Considerations
- Authentication tasks are P0 (blocks most features)
- GDPR compliance tasks are high priority (legal risk)
- User journey: Application flow > Discovery flow (core value)
- External API tasks: start early (dependency risk)
Integration Points
- BMAD Scrum Master: Sprint planning coordination
- Ralph Wiggum: /ralph-loop for autonomous task execution
- All Agents: Receive prioritized task assignments
- Test Manager: Test priority aligned with implementation
Examples
Priority Queue - Sprint 3
P0 - Critical (Do Now):
├─ TASK-004-01: OAuth2 implementation [Blocks: 12 tasks]
└─ TASK-010-03: Fix login security vulnerability
P1 - High (This Sprint):
├─ TASK-005-01: User profile API [Blocks: 4 tasks]
├─ TASK-005-03: Profile database migration
├─ TASK-006-01: University search API
└─ TASK-012-01: GDPR consent mechanism
P2 - Normal (If Capacity):
├─ TASK-007-02: Email notification service
├─ TASK-008-01: Application status dashboard
└─ TASK-009-01: Document upload feature
P3 - Backlog:
├─ TASK-015-01: Dark mode UI
└─ TASK-016-01: Analytics integration
Validation
- All tasks have priority assigned
- No circular priority dependencies
- Critical path tasks are P0/P1
- Agent workload balanced
- Sprint capacity not exceeded
Weekly Installs
1
Repository
vihang-hub/inte…rameworkGitHub Stars
1
First Seen
4 days ago
Security Audits
Installed on
amp1
cline1
openclaw1
opencode1
cursor1
kimi-cli1