Task Breakdown Framework
Purpose: Use this file during MAP to define hierarchy, size work, estimate effort, and stop decomposition at the right granularity.
Contents
- Task hierarchy
- Epic input template
- T-shirt sizing
- Complexity factors
- Estimation formula
- Breakdown rules
Task Hierarchy
| Level | Size | Description | Example |
|---|---|---|---|
| Epic | 1-5 days |
large feature or initiative | “Implement payment system” |
| Story | 2-8 hours |
user-facing slice | “Add checkout form” |
| Task | 30-120 min |
technical work unit | “Create PaymentForm component” |
| Atomic Step | 5-15 min |
single testable, committable action | “Define PaymentProps interface” |
Epic Input Template
## Epic: [Name]
**Goal**: [What we are trying to achieve]
**Success Criteria**: [How we know it is done]
**Constraints**: [Time, tech, scope limits]
**Out of Scope**: [What we are not doing]
**Initial Estimate**: [XS/S/M/L/XL]
**Risk Level**: [Low / Medium / High]T-Shirt Sizing
| Size | Minutes | Interpretation |
|---|---|---|
| XS | 5-10 |
trivial, clear path, no unknowns |
| S | 10-15 |
simple and ready to execute |
| M | 15-30 |
too large for an atomic step, split further |
| L | 30-60 |
complex task, must be decomposed |
| XL | 60+ |
always decompose immediately |
Complexity Factors
| Factor | Multiplier | Meaning |
|---|---|---|
| New technology | 1.5x |
first time using a library or API |
| Unclear requirements | 1.5x |
investigation still needed |
| External dependency | 2.0x |
approval, third party, or cross-team dependency |
| High risk | 1.5x |
could break existing functionality |
| Multiple files | 1.3x |
changes span several modules |
Estimation Formula
Actual Time = Base Estimate × Complexity Multiplier × Risk Buffer
Risk Buffer:
- Low: 1.0x
- Medium: 1.3x
- High: 1.5xEstimation Output
### Time Estimate: [Step Name]
| Aspect | Value |
|--------|-------|
| Base Size | M (20 min) |
| Complexity | New API (1.5x) |
| Risk Level | Medium (1.3x) |
| Estimated | 39 min |
Over the 15-minute threshold -> split further.Breakdown Rules
- Keep breaking work down until the current step is under
15 min. - Every atomic step must be testable.
- Every atomic step must be committable.
- Identify the responsible agent for each step.
- mark dependencies before execution starts.
- Flag XL items immediately.
Sources
- DORA 2025 report — small, frequent commits correlate with higher software delivery performance. AI-generated PRs are 51% larger on average; atomic decomposition counteracts cognitive-overload and review-skip risk. dora.dev/research/2025/dora-report/
- SAFe 6.0 Story hierarchy and "Reimagining SAFe" (2024-2025 updates) — Epic → Feature → Story → Task granularity. framework.scaledagile.com/whats-new-in-safe-6-0/
- Linear Agent (March 2026) — AI-generated issue hierarchies as raw MAP-phase input for Sherpa validation. linear.app/changelog/2026-03-24-introducing-linear-agent