---
name: a5c-ai/work-decomposition
source: https://app.decimal.ai/s/a5c-ai-work-decomposition@1/SKILL.md
source_sha256: 1f7cefc5f7f3
---

- When a goal is too large for a single agent
- When parallel execution would benefit progress
- When work needs tracked attribution

## Process

1. **Analyze** the goal and project context
2. **Identify** natural seams for decomposition
3. **Create MEOWs** with clear boundaries and dependencies
4. **Classify** as beads (persistent) or wisps (ephemeral)
5. **Map dependencies** between MEOWs
6. **Estimate** effort and assign priorities

## Decomposition Principles

- Each MEOW should be completable by a single agent
- Dependencies should form a DAG (no cycles)
- Prefer more smaller beads over fewer larger ones
- Wisps for throwaway work (scaffolding, exploration)
- Every MEOW gets attribution tracking

## Tool Use

Invoke via babysitter process: `methodologies/gastown/gastown-orchestrator` (analyze-work step)