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.
Epic Breakdown Advisor is an interactive product-management skill that helps split large epics into sprint-sized user stories using Richard Lawrence's Humanizing Work methodology. It solves the common backlog problem of items that are too big to estimate, sequence, or safely deliver, and it enforces vertical slicing (end-to-end user value across architectural layers) instead of horizontal slicing by technical component.
The workflow is a systematic, flowchart-driven three-step process: a pre-split INVEST validation (Independent, Negotiable, Valuable, Estimable, Testable, deliberately excluding "Small"), then sequential application of nine splitting patterns (workflow steps, CRUD operations, business-rule variations, data variations, data-entry methods, major effort, simple/complex, defer performance, and break out a spike) until one fits, and finally an evaluation step that favors splits which reveal low-value work to eliminate or which produce equal-sized stories. A meta-pattern (identify core complexity, list variations, reduce to one complete slice, make the rest separate stories) runs across all patterns. The skill defers to a companion workshop-facilitation protocol for interaction style, asking one question per turn with numbered quick-select options and progress labels.
It targets product managers, product owners, and agile teams preparing for sprint planning or backlog refinement. Typical use is pasting an epic from Jira or Linear and being walked through validation and pattern selection to arrive at smaller, value-preserving stories.
Provide the epic or large story (paste it as written from your backlog, plus optional team context like sprint length). The advisor runs an INVEST check, then walks the nine splitting patterns in order against your epic.
Workflow steps, operations (CRUD), business-rule variations, data variations, data-entry methods, major effort, simple/complex, defer performance, and break out a spike, applied sequentially until one fits.
Because the whole point of the exercise is that the item is currently too large. It validates the other five criteria before splitting, and stops if the story fails 'Valuable' (a technical task should be reframed or combined, not split).
It references a companion workshop-facilitation skill as its interaction protocol for session structure, one-question turns, and numbered response options. Its own file holds the domain-specific splitting logic.
None beyond the conversation. It is a single-file interactive advisor and works with pasted epic text from tools like Jira or Linear.
Quick Setup:
.claude/skills/Repository
deanpeters/product-manager-skills