What is progressive disclosure in Claude engineering?

It is a design pattern where information is revealed in stages (metadata, then instructions, then full resources) to minimize token costs and maximize context efficiency.

How do hooks work in Claude Code?

Hooks provide a universal event system to validate, modify, or log actions before (PreToolUse) or after (PostToolUse) a tool is executed by the agent.

Is this skill compatible with the Claude Dev Agent SDK?

Yes, it includes specific patterns, reference paths, and navigation guides for managing sessions, subagents, and tool-use within the Claude Dev platform.

Why distinguish between stable and volatile information?

Stable principles like architecture patterns rarely change, while volatile details like API field names change frequently; this skill provides workflows to fetch the latest official docs for volatile data.

Claude Engineering Best Practices

Name: Claude Engineering Best Practices
Author: Git-Fg

byGit-Fg

•

学習とドキュメンテーション

Implements standardized architecture and security patterns for Claude Code plugins and Claude Dev agent development.

概要

This skill serves as a comprehensive engineering framework for developers building on the Claude platform, providing high-level principles like progressive disclosure and hook-based event systems alongside volatile technical lookups. It bridges the gap between conceptual architecture and practical implementation by offering workflows for plugin development, agent SDK configuration, and security-first sandboxing. Whether you are designing a new skill or auditing an existing agent, this tool ensures alignment with official Claude engineering standards for performance and reliability.

主な機能

Three-tier skill authoring workflows (Design, Implement, Test)
Hook-based event systems for PreToolUse and PostToolUse validation
Progressive disclosure architecture for optimized token usage
1 GitHub stars
Security-first design patterns including sandboxing and IAM least-privilege
Dynamic lookup workflows to verify volatile API details and schemas

ユースケース

Developing custom Claude Code plugins with standardized manifests and caching
Configuring Claude Dev Agent SDKs with robust session persistence and subagents
Auditing AI agents for security compliance and engineering quality

概要

主な機能

Three-tier skill authoring workflows (Design, Implement, Test)
Hook-based event systems for PreToolUse and PostToolUse validation
Progressive disclosure architecture for optimized token usage
1 GitHub stars
Security-first design patterns including sandboxing and IAM least-privilege
Dynamic lookup workflows to verify volatile API details and schemas

ユースケース

Developing custom Claude Code plugins with standardized manifests and caching
Configuring Claude Dev Agent SDKs with robust session persistence and subagents
Auditing AI agents for security compliance and engineering quality