Skip to main content

Initiative System

The initiative system is a strategic planning layer above epics that enables:

  • Goal-oriented work: Epics serve initiatives, not just tasks
  • Success criteria: Measurable/verifiable conditions for "achieved"
  • Sufficiency checking: Post-epic QA to confirm work met goals
  • Autonomous operation: Agents can pick up work and verify completion

Mental Model

Initiative (strategic goal)
├── Criteria (what "achieved" means)
│ ├── Criterion 1: "Users can export data"
│ ├── Criterion 2: "API response < 200ms"
│ └── Criterion 3: "Test coverage > 80%"

└── Epics (work to achieve the goal)
├── Epic 1 → Review → "sufficient" (addressed criteria 1)
├── Epic 2 → Review → "partial" (mostly addressed 2)
└── Epic 3 → Review → ... (continue until all criteria met)

Initiative Lifecycle

draft → active → achieved

paused → active

abandoned
StatusMeaning
draftPlanning phase, not ready for work
activeWork is happening against this initiative
pausedTemporarily suspended (blocked, deprioritized)
achievedAll success criteria verified as met
abandonedIntentionally stopped (no longer relevant)

Success Criteria

Criteria define what "achieved" means. Each criterion has:

  • description: What needs to be true
  • verification_method: How to check (manual, automated, metric)
  • For metrics: metric_name, metric_target, metric_operator

Good Criteria Examples

✓ "Agent can create a devlog entry with only service_id, title, and content"
✓ "Response time p95 < 200ms under 100 concurrent users"
✓ "All public APIs have OpenAPI documentation"
✓ "New agent can complete onboarding checklist without human help"

Poor Criteria (Too Vague)

✗ "System is fast"
✗ "Code is clean"
✗ "Users are happy"

Epic Reviews

When an epic completes, create a review to assess sufficiency:

OutcomeMeaningNext Step
sufficientFully addressed relevant criteriaMark criteria achieved
partialMade progress but gaps remainDocument gaps, may need follow-up
insufficientDidn't meaningfully advance criteriaAnalyze why, create corrective epic

Workflow Patterns

Starting an Initiative

# 1. Create the initiative
initiative = mcpammer_initiative_create(
title="User Export Feature",
description="Enable users to export their data in multiple formats",
rationale="GDPR compliance and user empowerment",
priority="high"
)

# 2. Add success criteria
mcpammer_initiative_add_criterion(
initiative_id=initiative.id,
description="Users can export to CSV, JSON, and PDF",
verification_method="manual"
)

# 3. Activate when ready
mcpammer_initiative_update(initiative_id=initiative.id, status="active")

Working on Initiative Epics

# 1. Find next work
work = mcpammer_next_work(assignee="claude-code")

# 2. Create epic linked to initiative
epic = mcpammer_ticket_create(
title="Implement CSV Export",
ticket_type="epic"
)
mcpammer_initiative_link_epic(initiative_id=initiative.id, epic_id=epic.id)

# 3. Do the work, complete the epic
mcpammer_ticket_complete(ticket_id=epic.id)

# 4. Review the completed epic
mcpammer_epic_review_create(
epic_id=epic.id,
outcome="partial",
summary="CSV export working, JSON and PDF still needed",
gaps_identified="JSON and PDF formats not yet implemented"
)

Checking Progress

# Get full initiative details
initiative = mcpammer_initiative_get(initiative_id=id)
# Returns: epics_count, epics_done, criteria_count, criteria_achieved, progress_percent

# Check if we've achieved the goal
result = mcpammer_initiative_check_achievement(initiative_id=id)
if result.achieved:
print("Initiative complete!")
else:
print(f"Remaining: {result.remaining}")

Key Principles

  1. Initiatives are goals, not projects - Focus on outcomes, not activities
  2. Criteria should be verifiable - If you can't check it, you can't achieve it
  3. Reviews are honest assessments - "Partial" and "insufficient" are valid outcomes
  4. Work serves initiatives - Every epic should advance some criterion
  5. Achievement is automatic - When all criteria are met, the system knows