How do I ensure a skill is token-efficient?

Move heavy details to tool help commands, use cross-references instead of repeating instructions, and target word counts under 200 for frequently-loaded skills.

What is Claude Search Optimization (CSO)?

CSO is a set of practices including rich keyword coverage and trigger-based descriptions that help Claude correctly identify and load the right skill for a specific task.

What is the Red-Green-Refactor cycle for AI skills?

It involves creating a pressure scenario (test), observing the agent fail (Red), writing the skill documentation (Green), and then closing logic loopholes while maintaining compliance (Refactor).

Why shouldn't I summarize the skill workflow in the description?

If a description summarizes the process, Claude may follow the summary as a shortcut and skip reading the detailed instructions in the full skill file.

Where are Claude Code skills stored locally?

Personal skills are typically stored in agent-specific directories, such as ~/.claude/skills for Claude Code.

Claude Code Skill Architect

Name: Claude Code Skill Architect
Author: lastdays03

bylastdays03

Learning & Documentation

Applies Test-Driven Development principles to create, verify, and optimize specialized AI capabilities for Claude Code.

About

The writing-skills plugin provides a rigorous framework for developing Claude Code skills using a Test-Driven Development (TDD) approach. It treats documentation as production code and agent behavior as the test surface, guiding users through the Red-Green-Refactor cycle to ensure every skill effectively modifies agent behavior. By focusing on triggering conditions rather than workflow summaries, it maximizes Claude Search Optimization (CSO) and ensures future Claude instances discover and apply the correct techniques with minimal token overhead.

Key Features

Standardized SKILL.md templates with YAML frontmatter guidelines
Graphviz and flowchart integration standards for complex decision logic
Claude Search Optimization (CSO) patterns for improved skill discovery
Token-efficiency strategies for high-frequency background skills
TDD-based skill creation workflow (Red-Green-Refactor for documentation)
0 GitHub stars

Use Cases

Verifying that documentation actually corrects agent logic through pressure scenarios
Developing new specialized domain capabilities for Claude Code
Optimizing skill discovery and reducing context window bloat

About

Key Features

Standardized SKILL.md templates with YAML frontmatter guidelines
Graphviz and flowchart integration standards for complex decision logic
Claude Search Optimization (CSO) patterns for improved skill discovery
Token-efficiency strategies for high-frequency background skills
TDD-based skill creation workflow (Red-Green-Refactor for documentation)
0 GitHub stars

Use Cases

Verifying that documentation actually corrects agent logic through pressure scenarios
Developing new specialized domain capabilities for Claude Code
Optimizing skill discovery and reducing context window bloat