Initiative skill: planning, status, and performance

Comprehensive initiative management with three distinct modes: planning new initiatives, tracking status, and generating performance reports with KPI analysis.

Explicitly excludes

This does NOT provide: hour estimates, sprint planning, developer task breakdowns, ticket creation, complexity points.

Planning mode

Scope new initiatives, build business cases, and assess project feasibility from a product management perspective.

1. Skill sets required

Identify competencies needed:

For each skill note:

• Proficiency level (Basic / Intermediate / Expert)
• Coverage (Covered / Gap / Partial)

2. Milestone breakdown

Outcome-oriented milestones, not tasks:

3. Dependencies, risks, assumptions

• Dependencies: Internal teams, external vendors, data sources
• Risks: Technical, resource, timeline, market
• Assumptions: What must be true for success?

4. Complexity and effort (qualitative)

Effort profile: Light / Medium / Heavy / Major Door type: One-Way (irreversible) / Two-Way (reversible)

Status mode

When invoked for initiative status (not planning):

1. Read initiative from memory/initiatives/{name}.md
2. Query project tracker for current sprint status:
   • Open items with initiative label
   • Blocked items
   • Completion metrics
3. Query recent journal entries for meeting notes mentioning the initiative
4. Compile health report:
   • Status: On Track / At Risk / Blocked
   • Progress: Stories done vs total
   • Blockers: Open blocked items
   • Recent decisions: From memory/decisions/
   • Upcoming milestones: From initiative memory file
   • Recommendations: Based on data

Performance mode

Generate KPI-based performance reports with trend analysis, separate from initiative-status (which gives holistic project health). This mode focuses on operational metrics and trends.

Step 1: Read KPI definitions

Read reference/kpis.md to determine:

• Which teams/initiatives to report on
• Which metrics to pull
• Data sources for each metric

Step 2: Query data sources

Project tracker

Common query patterns per metric:

Time tracking (when configured)

• Time and utilization data
• Billable hours by team member

Monitoring (when configured)

• Pipeline reliability, uptime
• Infrastructure cost trends

Step 3: Calculate trends

For each metric:

• Current period: Last 2 weeks (or current sprint)
• Previous period: Prior 2 weeks (or previous sprint)
• Trend: Improving / Declining / Stable
  • Improving: current is better than previous by >10%
  • Declining: current is worse than previous by >10%
  • Stable: within 10% variance

Step 4: Flag issues

Flag metrics that:

• Crossed a negative threshold (e.g., defect rate >15%)
• Show declining trend for 2+ consecutive periods
• Have items blocked >5 days
• Are missing data (MCP query failed or connector not configured)

Step 5: Generate report

# Performance report: YYYY-MM-DD

## [Team/Initiative Name]

| KPI | Current | Previous | Trend |
|-----|---------|----------|-------|
| [Metric] | [Value] | [Value] | [Improving/Declining/Stable] |

### Flagged issues
- [Issue description with specific data]

### Recommended actions
- [Data-driven recommendation]

Repeat for each team/initiative in reference/kpis.md.

Step 6: Save to journal

Save to journal/YYYY-MM/YYYY-MM-DD-performance-report.md

---
date: YYYY-MM-DD
title: Performance Report
type: performance
teams: [Team names]
initiatives: [Initiative names]
---

Output format (planning mode)

## Initiative: [Name]

### Executive summary
[2-3 sentences]

### Skill sets required
| Skill | Level | Coverage |
|-------|-------|----------|

### Milestones
1. **[Phase]:** [Outcome] -- Timeline range

### Dependencies
- Internal: [[Team/Person]]
- External: [[Vendor]]

### Risks and assumptions
| Risk | Likelihood | Impact | Mitigation |
|------|------------|--------|------------|

### Effort profile
- Technical complexity: [L/M/H]
- Coordination: [L/M/H]
- Overall: [Light/Medium/Heavy/Major]
- Door type: [One-Way / Two-Way]

Output format (status mode)

## Initiative status: [Name]

**Overall health:** [On Track / At Risk / Blocked]

### Progress
| Metric | Value |
|--------|-------|
| Stories completed | X / Y |
| Blocked items | N |
| Sprint velocity | Z pts |

### Blockers
- [Blocker description with owner]

### Recent decisions
- [Decision from memory/decisions/]

### Upcoming milestones
- [Milestone with target date]

### Recommendations
- [Data-driven recommendation]

Output format (performance mode)

## Performance report: YYYY-MM-DD

### [Team/Initiative Name]

| KPI | Current | Previous | Trend | Status |
|-----|---------|----------|-------|--------|
| [Metric] | [Value] | [Value] | [Improving/Declining/Stable] | [OK/Warning/Critical] |

### Flagged issues
- [Issue description with specific data and timeline]

### Recommended actions
- [Data-driven recommendation with priority]

### Data source notes
- [Any limitations or gaps in metric collection]

Context budget

Planning mode:

• Memory: Read initiative file + up to 3 related files
• Project tracker: Up to 3 queries
• Journal: Current month _index.md + up to 2 relevant entries

Status mode:

• Memory: Read initiative file + up to 3 related files
• Project tracker: Up to 3 queries
• Journal: Current month _index.md + up to 2 relevant entries

Performance mode:

• Reference: reference/kpis.md (mandatory)
• Project tracker: Up to 3 queries per team/initiative
• Memory: Read initiative files for context (up to 3)
• Time tracking: Limited to configured integration
• Monitoring: Limited to configured integration

Absolute constraints

Universal constraints from the core skill apply (index-first pattern, journal persistence, integration constraints). Additionally:

Planning mode:

• NEVER provide hour or sprint estimates
• NEVER create tickets (report only)
• NEVER skip dependency analysis

Status mode:

• NEVER skip project tracker query (fall back to memory-only if not configured)

Performance mode:

• NEVER fabricate metric data (if query fails, report the gap)
• NEVER skip trend calculation
• NEVER output without the flagged issues section (even if empty)
• ALWAYS note data source limitations in the report