Epic Breakdown AdvisorSAFE

Product Management skills framework built on battle-tested methods for Claude Code, Cowork, Codex, and AI agents.

The instruction file, verbatim from the audited commit — this is the text the model reads, and the surface the audit's instruction layer examines. Quoted here so you can judge it without cloning anything.

---
name: epic-breakdown-advisor
argument-hint: "[epic to split]"
description: Break down epics into user stories with Humanizing Work split patterns. Use when a backlog item is too large to estimate, sequence, or deliver safely.
intent: >-
  Guide product managers through breaking down epics into user stories using Richard Lawrence's complete Humanizing Work methodology—a systematic, flowchart-driven approach that applies 9 splitting patterns sequentially. Use this to identify which pattern applies, split while preserving user value, and evaluate splits based on what they reveal about low-value work you can eliminate. This ensures vertical slicing (end-to-end value) rather than horizontal slicing (technical layers).
type: interactive
best_for:
  - "Splitting epics into smaller vertical slices"
  - "Choosing the right story split pattern for a large backlog item"
  - "Turning vague feature blobs into sprint-sized stories"
scenarios:
  - "Break this onboarding epic into smaller user stories"
  - "Help me split a large reporting feature before sprint planning"
  - "Which story-splitting pattern should I use for this admin workflow epic?"
theme: pm-artifacts
estimated_time: "20-30 min"
---


## Purpose
Guide product managers through breaking down epics into user stories using Richard Lawrence's complete Humanizing Work methodology—a systematic, flowchart-driven approach that applies 9 splitting patterns sequentially. Use this to identify which pattern applies, split while preserving user value, and evaluate splits based on what they reveal about low-value work you can eliminate. This ensures vertical slicing (end-to-end value) rather than horizontal slicing (technical layers).

This is not arbitrary slicing—it's a proven, methodical process that starts with validation, walks through patterns in order, and evaluates results strategically.

## Input

**Works best with:** The epic or large story you need to break down — paste it as written in your backlog.
**Also useful:** Team context (sprint length, estimation ceiling) and what's blocking delivery (too big to estimate, sequence, or release).

Anything supplied with the invocation itself — text after the skill name, a pasted context dump, or an appended `ARGUMENTS:` line — counts as answers already given. Use it and skip whatever it covers; don't re-ask.

**Arriving empty-handed? That works too.** The advisor opens by asking for the epic text, then walks the Humanizing Work pattern flowchart against it.

**Example invocation:** `Break this down: 'As a finance admin, I can export any report to Excel, PDF, or CSV with custom date ranges and saved templates.'`

## Key Concepts

### Core Principles: Vertical Slices Preserve Value
A user story is "a description of a change in system behavior from the perspective of a user." Splitting must maintain **vertical slices**—work that touches multiple architectural layers and delivers observable user value—not horizontal slices addressing single components (e.g., "front-end story" + "back-end story").

### The Three-Step Process
1. **Pre-Split Validation:** Check if story satisfies INVEST criteria (except "Small")
2. **Apply Splitting Patterns:** Work through 9 patterns sequentially until one fits
3. **Evaluate Splits:** Choose the split that reveals low-value work or produces equal-sized stories

### The 9 Splitting Patterns (In Order)
1. **Workflow Steps** — Thin end-to-end slices, not step-by-step
2. **Operations (CRUD)** — Create, Read, Update, Delete as separate stories
3. **Business Rule Variations** — Different rules = different stories
4. **Data Variations** — Different data types/structures
5. **Data Entry Methods** — Simple UI first, fancy UI later
6. **Major Effort** — "Implement one + add remaining"
7. **Simple/Complex** — Core simplest version first, variations later
8. **Defer Performance** — "Make it work" before "make it fast"
9. **Break Out a Spike** — Time-box investigation when uncertainty blocks splitting

### Meta-Pattern (Applies Across All Patterns)
1. Identify the core complexity
2. List all variations
3. Reduce variations to **one complete slice**
4. Make other variations separate stories

### Why This Works
- **Prevents arbitrary splitting:** Methodical checklist prevents guessing
- **Preserves user value:** Every story delivers observable value
- **Reveals waste:** Good splits expose low-value work you can deprioritize
- **Repeatable:** Apply to any epic consistently

---

### Facilitation Source of Truth

Use [`workshop-facilitation`](../workshop-facilitation/SKILL.md) as the default interaction protocol for this skill.

It defines:
- session heads-up + entry mode (Guided, Context dump, Best guess)
- one-question turns with plain-language prompts
- progress labels (for example, Context Qx/8 and Scoring Qx/5)
- interruption handling and pause/resume behavior
- numbered recommendations at decision points
- quick-select numbered response options for regular questions (include `Other (specify)` when useful)

This file defines the domain-specific assessment content. If there is a conflict, follow this file's domain logic.

## Application

### Step 0: Provide Epic Context

**Agent asks:**

Please share your epic:

- Epic title/ID
- Description or hypothesis
- Acceptance criteria (especially multiple "When/Then" pairs)
- Target persona
- Rough estimate

**You can paste from Jira, Linear, or describe briefly.**

---

### Step 1: Pre-Split Validation (INVEST Check)

**Before splitting, verify your story satisfies INVEST criteria (except "Small"):**

**Agent asks questions sequentially:**

**1. Independent?**
"Can this story be prioritized and developed without hard technical dependencies on other stories?"

**Options:**
- Yes — No blocking dependencies
- No — Requires other work first (flag this)

---

**2. Negotiable?**
"Does this story leave room for the team to discover implementation details collaboratively, rather than prescribing exact solutions?"

**Options:**
- Yes — It's a conv

SAFEgrade B · trust 89/100 Nothing in the source contradicts what it says it does. Grade A is reserved for packages that have also passed the behavioural sandbox.

What the source does

Filesystem

none-observed

Network

none-observed

Shell

none-observed

Dependencies

pinned

Secrets in source

none-found

Findings (0)

No findings outside the package's declared scope.

Gates applied: no_behavioural_pass.