Skip to content

digitalhand/claude-skills-agents

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

5 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

Godot, light-weight Everything-Claude-Code

A Godot C#, GDScript curated subset of the everything-claude-code repository. Strips out public-use rules and heavy JavaScript/frontend context, focusing Godot specific developers, and some other front-end, back-end support.

Table of Contents

Focus

  • Languages: Godot (GDScript + C#), C# (.NET / ASP.NET Core), Java (Spring Boot), Terraform (Azure Verified Modules), Go, Python
  • Cloud: Azure
  • Philosophy: No auto-pushing code at this level — override at project/user level if needed. Plugins are under review and not yet vetted.

Installation

Linux / macOS (Bash)

./install.sh <language> [<language> ...]

# Examples
./install.sh golang
./install.sh golang java terraform
./install.sh --all

Windows (PowerShell)

.\install.ps1 <language> [<language> ...]

# Examples
.\install.ps1 golang
.\install.ps1 golang java terraform
.\install.ps1 -All

Available Languages

Run the installer with no arguments to see the list:

  • csharp
  • godot
  • golang
  • java
  • terraform

Install All Languages

Use --all (Bash) or -All (PowerShell) to install every available language at once.

What Gets Installed

The installer copies the following into ~/.claude/:

What Source Destination
Common rules rules/common/ ~/.claude/rules/common/
Language rules rules/<lang>/ ~/.claude/rules/<lang>/
Agents agents/*.md ~/.claude/agents/
Skills skills/*/SKILL.md ~/.claude/skills/<name>/

Common rules and all agents/skills are always installed. Language rules are installed only for the languages you request (or all with --all).

Repository Structure

rules/
  common/       Language-agnostic rules (coding style, git, security, testing, etc.)
  csharp/       C#/.NET-specific rules
  godot/        Godot engine rules (GDScript + Godot C#)
  golang/       Go-specific rules
  java/         Java/Spring Boot-specific rules
  terraform/    Terraform/Azure-specific rules
agents/         Agent definitions (planner, architect, tdd-guide, code-reviewer, etc.)
skills/         Skill definitions (Spring Boot, Go, Python, JPA, Terraform, etc.)
examples/       Example CLAUDE.md files for target projects
plugins/        Plugin marketplace docs (under review, not yet vetted)
install.sh      Bash installer (Linux/macOS)
install.ps1     PowerShell installer (Windows)

Using Agents

How Agents Work

Agents are specialized sub-agents that Claude Code delegates to for specific tasks. Each agent is defined as a Markdown file in agents/ with YAML frontmatter specifying its name, description, allowed tools, and model tier.

After installation, agents live in ~/.claude/agents/ and Claude Code discovers them automatically. You do not need to configure anything beyond running the installer.

Invoking agents manually:

> Can you run the code-reviewer agent on my latest changes?
> Use the planner agent to break down this feature.
> Run security-reviewer before I commit.

Automatic invocation: The rules in rules/common/agents.md instruct Claude Code to invoke agents proactively based on context — for example, running code-reviewer after code modifications or tdd-guide when writing new features. You do not need to ask explicitly; Claude Code will use them when appropriate.

Parallel execution: Independent agents run in parallel when possible. For example, security-reviewer and code-reviewer can analyze different aspects of the same change simultaneously.

Available Agents

Agent Model Purpose When to Use
planner opus Implementation planning Complex features, architectural changes
architect opus System design and scalability Architectural decisions, technical trade-offs
tdd-guide sonnet TDD enforcement New features, bug fixes, refactoring
code-reviewer sonnet Code quality and security Immediately after writing or modifying code
security-reviewer sonnet Vulnerability detection (OWASP Top 10) Code handling user input, auth, secrets, APIs
build-error-resolver sonnet Fix build/type errors When builds fail
refactor-cleaner sonnet Dead code cleanup Code maintenance, consolidation

When Agents Are Invoked

Trigger Agent
User requests a complex feature or refactoring planner
Architectural decision or trade-off discussion architect
New feature, bug fix, or any code change tdd-guide
Code has just been written or modified code-reviewer
Code touches auth, user input, secrets, or APIs security-reviewer
Build or compilation fails build-error-resolver
Code maintenance or dead code cleanup refactor-cleaner

Skills

Godot

Skill Description
godot-patterns File formats (.gd, .tscn, .tres), component architecture, signals, CLI, common pitfalls
godot-csharp-patterns Godot 4 C# node lifecycle, signals, scenes, input, [GlobalClass], RefCounted, interop
godot-gdscript-patterns GDScript static typing, annotations, signals, coroutines, style conventions
godot-testing GUT/gdUnit4 for GDScript, NUnit for C#, headless execution, CI/CD, benchmarking

C# / .NET

Skill Description
csharp-coding-standards Naming, immutability, nullable refs, modern C# features
csharp-patterns ASP.NET Core architecture, DI, middleware, caching
csharp-testing xUnit, Moq, NSubstitute, FluentAssertions, Testcontainers
aspnetcore-security JWT auth, authorization, validation, CORS, rate limiting
efcore-patterns Entity design, queries, N+1 prevention, migrations, auditing

Java / Spring Boot

Skill Description
java-coding-standards Naming, immutability, Optional, streams, exceptions
springboot-patterns Architecture, REST API, layered services, caching
springboot-security Auth, validation, CSRF, secrets, headers
springboot-tdd JUnit 5, Mockito, MockMvc, Testcontainers, JaCoCo
springboot-verification Build, static analysis, tests, security scans
jpa-patterns Entity design, relationships, queries, transactions

Terraform

Skill Description
terraform-style-guide HCL style conventions and best practices
terraform-patterns Azure Verified Modules, resource patterns

Go

Skill Description
golang-patterns Idioms, concurrency, error handling, packages
golang-testing Table-driven tests, benchmarks, fuzzing, coverage

Python

Skill Description
python-patterns PEP 8, type hints, idioms, best practices
python-testing pytest, TDD, fixtures, mocking, parametrization

Cross-Language

Skill Description
api-design REST resource naming, status codes, pagination, errors
security-review Security checklist and patterns
tdd-workflow TDD enforcement with 80%+ coverage
verification-loop Comprehensive verification for Claude Code sessions
postgres-patterns Query optimization, schema design, indexing
agent-harness-construction AI agent action spaces, tool definitions
frontend-patterns React, TypeScript, state management

Using Rules

How Rules Work

Rules are Markdown files that Claude Code loads automatically to guide its behavior. They define coding standards, security requirements, workflows, and conventions that Claude Code follows when working in your project.

After installation, rules live in ~/.claude/rules/. Claude Code loads them based on two mechanisms:

  1. Common rules (rules/common/) — always loaded for every session, regardless of what files you are editing. These define universal standards like git workflow, security checks, and testing requirements.

  2. Language-specific rules (rules/<lang>/) — loaded conditionally based on the paths: globs in their YAML frontmatter. For example, Java rules specify:

    ---
    paths:
      - "**/*.java"
      - "**/pom.xml"
    ---

    These rules activate only when Claude Code is working with files matching those patterns. This keeps context focused — Terraform rules do not load when you are editing Java files, and vice versa.

You do not invoke rules manually. They are loaded automatically and shape how Claude Code writes code, reviews changes, handles errors, and interacts with your project. Think of them as persistent configuration, not commands.

Customizing rules: To override or extend rules for a specific project, add project-level rules in your project's .claude/rules/ directory or adjust behavior in your project's CLAUDE.md. Project-level settings take precedence.

Common Rules

Applied to all projects regardless of language:

Rule Purpose
coding-style Immutability, file organization (200-400 lines typical, 800 max), error handling
git-workflow Conventional commits (feat:, fix:, refactor:, etc.), PR workflow
development-workflow Research/reuse first, plan, TDD, code review, commit pipeline
security Mandatory pre-commit security checks, secret management
testing TDD workflow, 80% minimum coverage, unit/integration/E2E tests
performance Model selection strategy (Haiku/Sonnet/Opus), context window management
hooks PreToolUse/PostToolUse hook patterns, TodoWrite best practices
agents Agent orchestration, parallel execution, multi-perspective analysis

Language-Specific Rules

Each language directory contains 5 rule files that extend the common rules with language-specific guidance:

Rule Purpose
coding-style Language-specific formatting, naming conventions, idioms
patterns Architecture and design patterns (e.g., Spring Boot layers, Go interfaces)
security Language-specific security concerns (e.g., Spring Security, Go crypto)
testing Frameworks, tools, coverage configuration (e.g., JUnit 5, go test)
hooks Post-edit automation — formatters, compilers, linters

Available language rule sets: csharp, godot, golang, java, terraform.

Adding New Content

  • Language rules — add a new directory under rules/<language>/ with .md files. The installer picks it up automatically.
  • Skills — add skills/<skill-name>/SKILL.md.
  • Agents — add agents/<name>.md.

Key Conventions

  • Git: Conventional commits (feat:, fix:, refactor:, etc.). No PRs created at this level.
  • Workflow: Research/reuse first, plan (planner agent), TDD (tdd-guide agent), code review (code-reviewer agent), commit.
  • Code style: Immutability preferred. 200-400 lines typical, 800 max. Many small files over few large ones.
  • Testing: TDD mandatory, 80% minimum coverage.
  • Agents: Invoked proactively — planner for complex features, tdd-guide for new code, code-reviewer after writing. Parallel execution for independent tasks.

About

Claude agents, skills, and rules for C#, Godot, and other related languages.

Topics

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

 
 
 

Contributors