What are the core principles of the ref-development-standards skill?

The skill focuses on five key areas: asking the user when uncertain, researching best practices before coding, using specific tools for file operations, avoiding document proliferation, and requiring approval for new dependencies.

When should I activate this skill in my workflow?

You should activate it at the start of any development task, including feature implementation, architectural changes, tool selection, or when creating new documentation.

Does this skill work with other agents?

Yes, it is designed to delegate tasks like deep web research to the 'research-best-practices' agent while maintaining the overall development workflow and standards.

How does this skill improve AI coding results?

By providing a strict framework of rules and tool-usage patterns, it reduces 'creative' hallucinations and prevents the AI from making unauthorized or non-standard changes to a codebase.

Plan-Marshall Development Standards

Name: Plan-Marshall Development Standards
Author: cuioss

bycuioss

開発者ツール

Enforces consistent development principles, tool usage patterns, and decision-making workflows to ensure predictable AI-driven coding outputs.

概要

This skill serves as the foundational governance layer for the Plan-Marshall ecosystem, providing a set of standardized rules that ensure Claude follows best practices throughout the software development lifecycle. It guides the AI on critical decision-making points—such as when to consult the user versus proceeding independently—mandates research via specialized agents before implementation, and enforces the use of specific file-manipulation tools over generic bash commands. By centering development around these core standards, the skill minimizes architectural drift, prevents unauthorized dependency bloat, and maintains a high level of project integrity and reliability.

主な機能

Interactive Decision Tree: Determines when to ask for user input versus proceeding with autonomous tasks.
Standardized Tool Patterns: Enforces the use of high-level Read, Write, and Edit tools for better file system safety.
0 GitHub stars
Document Control: Minimizes file proliferation by prioritizing the reuse of existing context-relevant documentation.
Dependency Governance: Prevents unauthorized library or package additions without explicit user approval.
Research Integration: Mandates the discovery of current best practices via specialized agents before code generation.

ユースケース

Standardizing file operations and search patterns to prevent errors during complex refactoring tasks.
Ensuring all AI-generated code adheres to current industry best practices through automated research workflows.
Establishing a consistent architectural framework when starting new features or projects in Claude Code.

概要

主な機能

Interactive Decision Tree: Determines when to ask for user input versus proceeding with autonomous tasks.
Standardized Tool Patterns: Enforces the use of high-level Read, Write, and Edit tools for better file system safety.
0 GitHub stars
Document Control: Minimizes file proliferation by prioritizing the reuse of existing context-relevant documentation.
Dependency Governance: Prevents unauthorized library or package additions without explicit user approval.
Research Integration: Mandates the discovery of current best practices via specialized agents before code generation.

ユースケース

Standardizing file operations and search patterns to prevent errors during complex refactoring tasks.
Ensuring all AI-generated code adheres to current industry best practices through automated research workflows.
Establishing a consistent architectural framework when starting new features or projects in Claude Code.