Overview

Story Decomposition breaks complex specs into small, independently implementable stories with clear dependencies. It estimates effort and creates a dispatching queue so multiple coders can work in parallel.

How This Skill Works

The Architect reads the technical spec and existing code to identify implementable stories. It uses Grep/Glob to map touchpoints and interfaces, then writes formal story definitions and refines them with Edit. Each story is scoped to be completed by a single coder in one batch, with explicit dependencies so outputs can feed a parallel coder dispatch.

When to Use It

• When a large spec is too big for a single task and needs decomposition
• When feature dependencies are complex and require explicit ordering
• When you must maximize parallel work through a dispatch queue
• When each story must be completed by a single coder in one batch
• When integrating with maestro orchestration (phase 3) for story decomposition and prioritization

Quick Start

1. Step 1: Read the technical spec and use Grep/Glob to map touched modules
2. Step 2: Write initial story definitions and define dependency edges with Write
3. Step 3: Edit and validate each story so a single coder can complete it in one batch

Best Practices

• Break the spec into atomic stories with clear acceptance criteria
• Explicitly map and order dependencies to reveal the critical path
• Ensure each story is independently implementable by one coder in one batch
• Keep story scope minimal by isolating touched interfaces and modules
• Leverage Read/Grep/Glob and Write/Edit steps to iterate on story definitions

Example Use Cases

• Decompose a payment workflow into authorize, capture, and settlement stories
• Split a REST API upgrade into versioned, backward-compatible stories
• Break a data export feature into fetch, transform, and deliver stories
• Isolate a feature-flag rollout into story chunks with dependency order
• Parallelize a module refactor by exposing stable interfaces first

Frequently Asked Questions