decompose

Splits epics and oversized stories into independently deliverable slices recorded in a local backlog.

Updated Dec 1, 2025
One-click install
npx skills add https://github.com/jterratsdev/smart-deployment --skill decompose-jterratsdev
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: decompose
Source: https://github.com/jterratsdev/smart-deployment/tree/main/.setup-agents/skills/decompose
Command: npx skills add https://github.com/jterratsdev/smart-deployment --skill decompose-jterratsdev

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? Oversized epics and XXL stories cannot be estimated, planned, or delivered within a single iteration, and teams lack a consistent method to split them into valuable, testable slices while preserving traceability. ## Core Features & Use Cases - Split Gate Enforcement: Treats XXL sizing as a mandatory split gate and re-splits any slice that remains XXL before estimation. - Five Splitting Patterns: Splits by workflow steps, business rules, data variations, CRUD operations, or interfaces, keeping every slice INVEST-compliant. - Traceability Preservation: Links every slice to its parent epic or story, carries acceptance criteria down to the owning slice, and declares cross-slice dependencies. - Use Case: During backlog refinement, a BA finds an XXL story spanning three personas; the Skill produces a slice table (ID, title, parent, value, dependencies, size) and persists each slice as a local task via sf setup-agents task create. ## Quick Start Ask the assistant to decompose the XXL story about checkout into independently deliverable slices and record each one in the local backlog.

Frequently Asked Questions about decompose

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I split an epic into smaller stories?

Split an epic using five patterns: workflow steps, business rules, data variations, CRUD operations, or interfaces. Each resulting slice must be independently valuable and testable, and any slice still sized XXL is split again before estimation.

What does XXL story size mean in backlog refinement?

XXL is treated as a split gate, not an estimate. A story sized XXL must be decomposed into smaller slices before it can be estimated or planned, because it spans more than one iteration, persona, or bounded context.

How do I keep traceability when splitting a story?

Link every slice back to its parent epic or story ID, carry the parent's acceptance criteria down to the slice they belong to, and declare cross-slice dependencies explicitly so no acceptance criteria become orphaned.

Where are decomposed slices stored?

Slices are stored in the repo-resident backlog at .setup-agents/state/tasks.jsonl, created with the sf setup-agents task create command. They are not created in Jira; Atlassian MCP integration is optional and mirror-only.

When should I not split a story by technical layers?

Avoid splitting into technical layers such as frontend, backend, and database, because those slices deliver no standalone value. Each slice must remain independently valuable and testable according to INVEST principles.