How does this skill improve parallel execution?

The skill analyzes four types of dependencies (data, control, resource, and evidence) to identify tasks that don't overlap, allowing Claude to execute them simultaneously using multiple task tools.

What is an Atomic Task in ELD Sense?

An Atomic Task is the smallest unit of work in this framework, designed to be completed in 5-10 minutes. It focuses on a single conceptual change, is independently testable, and results in a single commit.

Can I use this for existing large repositories?

Yes, it is specifically designed for large tasks where implementation planning, decomposition, and parallelization are necessary to maintain project integrity and speed.

What is 'Context Inheritance' in this planning skill?

It is a strategy to pass only the minimum necessary information from a parent task to a child task. This prevents the AI from becoming confused by irrelevant data and optimizes token usage.

ELD Sense: Planning & Parallelization

Name: ELD Sense: Planning & Parallelization
Author: CAPHTECH

byCAPHTECH

0•

生产力与工作流

Decomposes complex coding tasks into atomic units and optimizes parallel execution with structured context inheritance.

ELD Sense Planning is a specialized capability for Claude designed to transform high-level implementation goals into a structured, three-level task hierarchy. By breaking down projects into 'Atomic Tasks'—defined as 5-10 minute units of work—it ensures every change is single-purpose, testable, and verifiable. The skill excels at managing context inheritance between parent and child tasks to prevent scope bloat, and it utilizes sophisticated dependency analysis to enable parallel execution via Claude's task tools. This approach maximizes development throughput while maintaining high quality through the integrated Evidence Ladder framework.

主要功能

01Evidence-based verification linked to specific project 'Laws' and 'Terms'

02Three-level hierarchical task decomposition (Root, Major, Atomic)

03Structured context inheritance to minimize noise and prevent scope creep

04Dependency-aware parallel execution optimization for task tools

05Atomic task standardization for 5-10 minute implementation cycles

060 GitHub stars

使用场景

01Breaking down large-scale feature implementations into manageable, parallel steps

02Optimizing development speed by identifying non-dependent tasks for simultaneous execution

03Designing complex system architectures with clear context boundaries between modules

主要功能

01Evidence-based verification linked to specific project 'Laws' and 'Terms'

02Three-level hierarchical task decomposition (Root, Major, Atomic)

03Structured context inheritance to minimize noise and prevent scope creep

04Dependency-aware parallel execution optimization for task tools

05Atomic task standardization for 5-10 minute implementation cycles

060 GitHub stars

使用场景

01Breaking down large-scale feature implementations into manageable, parallel steps

02Optimizing development speed by identifying non-dependent tasks for simultaneous execution

03Designing complex system architectures with clear context boundaries between modules