Why does this skill require a dedicated worktree?

Operating in a dedicated worktree (phase-5-task-3) ensures that the decomposition process is isolated, preventing conflicts with other development work and allowing for atomic commits and safe merge validation.

What are the required input files?

The primary prerequisite is a tasks.json file containing top-level tasks (typically 15-25) and a successful Phase 2 pipeline checkpoint.

What happens once the decomposition phase is finished?

The skill generates a completion signal (.signals/phase5-complete.json) and automatically triggers the Spec Generation phase (Phase 6) without requiring manual user intervention.

How does the Task Decomposer determine which tasks to expand?

The skill uses a scoring system from 0 to 10. Any task with a complexity score of 7 or higher is automatically flagged for research-backed expansion into detailed subtasks.

Can I process multiple tasks simultaneously?

Yes. The skill supports parallel subagent execution using Claude Code's task tool, allowing you to launch multiple agents to handle batches of tasks, which significantly reduces the total processing time.

Task Decomposer

Name: Task Decomposer
Author: Smith-Happens

bySmith-Happens

0•

Productividad y Flujo de Trabajo

Automates the breakdown of complex development tasks into actionable subtasks using scientific complexity scoring and isolated worktrees.

The Task Decomposer is a sophisticated workflow skill for Claude Code designed to manage Phase 5 of the development pipeline. It scientifically analyzes project tasks on a 0-10 complexity scale, automatically identifying high-effort items that require deeper granularization. For every complex task, it generates detailed subtasks complete with acceptance criteria and testing strategies. To ensure stability and performance, the skill utilizes dedicated worktree isolation to prevent merge conflicts and supports parallel subagent execution, which can accelerate the planning phase by up to 800% for large-scale projects.

Características Principales

01Scientific complexity scoring (0-10 scale) for objective task analysis

02Parallel subagent support for high-speed batch task expansion

03Autonomous pipeline transitions with automated completion signaling

04Automated research-backed subtask generation with acceptance criteria

05Isolated worktree execution (phase-5-task-3) for safe environment management

060 GitHub stars

Casos de Uso

01Scaling project planning by using parallel AI agents to expand large backlogs

02Breaking down monolithic development requirements into granular, testable subtasks

03Standardizing acceptance criteria and test strategies across complex engineering tasks

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add smith-happens/xlightsfpptester task-decomposer

For use in Claude.ai and ChatGPT

Download Skill