convez.ai
Skill
lelandg

.claude_code

lelandg/.claude_code

Ver no GitHub
InicialPluginScore · de 0 a 100

Claude Code config I'm currently using. Includes skills to merge into your workspace and document what you need.

Instalação

$npx convez-ai lelandg/.claude_code

1 comando no terminal: detecta o formato do repositório (marketplace ou plugin avulso) e instala via Claude Code. Requer Node 18+ e o Claude Code instalado.

Métricas

Estrelas
0
Forks
0
Acompanhando
0
Saúde
28%
Último commit
há 1 dia
Versão
—
Issues abertas
2
Contribuidores
1

README

A tradução para o português deste README ainda está na fila.

Claude Code Configuration

A comprehensive, production-tested configuration for Claude Code - Anthropic's CLI for Claude. This repository mirrors a real-world ~/.claude/ setup with custom agents, skills, output styles, and best-practice instructions.

Platform

This config is written for Linux (native Linux or WSL on Windows). macOS should work with minimal changes — swap python3 for python where needed and replace /home/<user> with /Users/<user> in permission entries.

Using Windows without WSL? Ask Claude to port it for you:

Here is my Claude Code config repo cloned at <repo-path>. Convert everything from Linux/WSL conventions to native Windows PowerShell equivalents: change shell scripts like statusline-command.sh to a PowerShell .ps1 (or keep Bash if using Git Bash), update /home/<USER>/ paths in claude/settings.json permissions to C:\Users\<USER>\ form, replace python3 with python, and adjust any bash ~/.claude/... invocations in settings.json / statusLine to whatever shell you want to use. Leave the rest (agents, skills, CLAUDE.md, instructions) alone — they are OS-neutral.

cl* skills — heads-up

Any skill prefixed with cl- (e.g. cl-project-init) is a sanitized example of a Chameleon Labs-specific workflow. The company-specific details (emails, product names, brand colors) have been replaced with YourCompany / yourdomain.com placeholders, but the skill is still structured around the author's use case. Treat it as a template you fork and rename for your own company, not something to install verbatim.

What's Included

.claude_code/
├── README.md
├── .claude/
│   └── commands/
│       ├── install-claude-config.md  # Bootstrap command (run after cloning)
│       └── yt-transcript.md          # /yt-transcript slash command
├── claude/                            # Mirrors ~/.claude/ — install this
│   ├── CLAUDE.md                      # Global instructions (loaded every session)
│   ├── CLAUDE_CodeMap.md              # CodeMap specification (language-neutral)
│   ├── settings.json                  # Claude Code settings & permissions
│   ├── mcp.json                       # MCP server configurations
│   ├── agentic_prompt_template.md     # Template for agentic workflows
│   ├── statusline-command.sh          # Custom status line script
│   ├── agents/                        # Custom agent definitions
│   │   ├── README.md                  # Agent documentation
│   │   ├── CLAUDE.md                  # Agent-specific instructions
│   │   ├── Claude-Code-Agents-Documentation.md  # Full agent architecture reference
│   │   ├── code-reviewer.md           # Code review agent (Opus)
│   │   ├── documentation-specialist.md# Documentation agent (Sonnet)
│   │   ├── performance-optimizer.md   # Performance analysis agent (Opus)
│   │   ├── research-assistant.md      # Research agent (Sonnet)
│   │   ├── software-engineer.md       # Coding agent (Opus)
│   │   ├── test-generator.md          # Test generation agent (Sonnet)
│   │   └── specs/
│   │       └── CLAUDE_CodeMap.md      # CodeMap spec (copy for agent access)
│   ├── instructions/                  # On-demand reference docs
│   │   ├── credentials.md             # Secrets management patterns
│   │   ├── file-operations.md         # File operation guidelines
│   │   └── plan-templates.md          # Implementation plan format
│   ├── output-styles/                 # Output formatting styles
│   │   ├── genui.md                   # Generative UI (HTML output)
│   │   ├── html.md                    # HTML/web development focus
│   │   └── technical-quality.md       # Comprehensive technical analysis
│   └── skills/                        # Custom skills (slash commands)
│       ├── time.md                    # Execution time tracking
│       ├── cl-project-init/           # Sanitized example: company project scaffolder
│       ├── claude-md-optimizer/       # CLAUDE.md optimization skill
│       ├── doc-all-projects/          # Sweep-and-regenerate docs across projects
│       ├── feature-documenter/        # Feature documentation skill
│       ├── graphify/                  # Any input → knowledge graph (HTML + JSON)
│       ├── install-claude-config/     # Install config into ~/.claude/
│       ├── product-manager/           # Product management toolkit
│       ├── project-documenter/        # Per-project user-facing docs generator
│       ├── raginclude-generator/      # Generate .raginclude file for RAG ingest
│       ├── sync-claude-config/        # Sync ~/.claude/ to this repo
│       ├── technical-documenter/      # Developer/support docs generator
│       ├── update-code-map/           # CodeMap maintenance skill
│       └── yt-transcript/             # Download YouTube transcripts
└── Docs/
    └── SETUP_GUIDE.md                 # Full setup & configuration guide

Installation

Prerequisites

  • Claude Code installed and working (claude --version)
  • Linux, WSL, or macOS shell (Bash/Zsh). Native Windows users should run the conversion prompt above first.
  • ~/.claude/ directory exists (created automatically by Claude Code on first run)
  • git installed

Option 1: Use the Install Command (Recommended)

Clone the repo, open Claude Code inside it, and run /install-claude-config.

git clone https://github.com/lelandg/.claude_code.git
cd .claude_code
claude .

Then in Claude Code:

/install-claude-config

The command is available immediately after cloning — no pre-installation needed.

What it does:

  1. Scans claude/ and compares every file against your existing ~/.claude/
  2. Presents a table: NEW, CHANGED, or IDENTICAL for each file
  3. For each changed file, shows a diff and asks: Install / Keep current / Smart merge
  4. Smart merge for settings.json: unions permissions.allow and enabledPlugins, keeps your personal values (defaultMode, trustedWorkspaces, etc.)
  5. Smart merge for CLAUDE.md: keeps your version, reports any new sections you don't have yet
  6. Backs up all overwritten files to ~/.claude/.backup/YYYY-MM-DD/ before writing
  7. Sets chmod +x on shell scripts automatically

Option 2: Copy Everything

git clone https://github.com/lelandg/.claude_code.git
cp -r .claude_code/claude/* ~/.claude/

Note: This overwrites existing files without merging. Back up ~/.claude/ first if you have existing config.


Option 3: Cherry-Pick What You Need

cd .claude_code

# Just the agents
cp claude/agents/*.md ~/.claude/agents/

# Just the skills
cp -r claude/skills/* ~/.claude/skills/

# Just the output styles
cp claude/output-styles/*.md ~/.claude/output-styles/

# Just the instructions
cp claude/instructions/*.md ~/.claude/instructions/

Option 4: Start from CLAUDE.md Only

cp .claude_code/claude/CLAUDE.md ~/.claude/CLAUDE.md

Customize it for your workflow. This single file gives you the core benefits: date handling, security rules, work procedures, and project conventions.


Required After Install

1. Personalize ~/.claude/CLAUDE.md — replace the placeholder table at the top:

| Your Name | your-github-username | your-discord-username |

2. Fix the Read permission path in ~/.claude/settings.json — find and update:

"Read(//home/<YOUR_USER>/.claude/**)"

Replace <YOUR_USER> with your actual username.

3. Set the GitHub token (required for GitHub MCP server):

# Add to ~/.bashrc or ~/.zshrc
export GITHUB_PERSONAL_ACCESS_TOKEN="ghp_your_token_here"

Generate a token at github.com/settings/tokens with repo, read:org, read:user scopes.

4. Restart Claude Code to load the new config.

See Docs/SETUP_GUIDE.md for the full guide including plugin installation, verification checklist, and troubleshooting.

Customization Guide

CLAUDE.md (Start Here)

The global CLAUDE.md is loaded into every Claude Code session. Customize these sections:

SectionWhat to Change
User & Contact InfoYour name, GitHub, Discord handles
Development EnvironmentsYour IDE, languages, tools
Runtime NotesYour OS, shell, virtual env setup
ScreenshotsYour screenshot storage path

settings.json

  • permissions.allow: Add Bash patterns for commands you frequently approve
  • permissions.defaultMode: "acceptEdits" auto-accepts file edits (recommended)
  • enabledPlugins: Toggle plugins on/off based on your needs
  • Read permissions: Update paths to match your home directory

mcp.json

  • github: Requires GITHUB_PERSONAL_ACCESS_TOKEN environment variable (setup instructions)
  • Add your own MCP servers as needed

Key Features

Agents

Six specialized agents for different tasks. Each has a focused system prompt, specified model (Opus for complex tasks, Sonnet for simpler ones), and tool access:

  • Code Reviewer - Thorough code review with verification-before-claims approach
  • Software Engineer - Full implementation capability with quality standards
  • Performance Optimizer - Algorithmic analysis, database optimization, caching strategies
  • Test Generator - Multi-framework test suite generation
  • Documentation Specialist - User and developer documentation
  • Research Assistant - Technology comparison and best practices research

Skills

Custom skills extend Claude Code with repeatable workflows:

  • update-code-map - Maintains a comprehensive CodeMap.md for codebase navigation
  • feature-documenter - Generates user-facing feature documentation from code analysis
  • project-documenter - Per-project user-facing feature docs + sitemap
  • technical-documenter - Developer and support-staff documentation (APIs, data models, errors)
  • doc-all-projects - Parallel sweep that regenerates stale docs across every registered project
  • claude-md-optimizer - Reduces CLAUDE.md token usage by extracting rarely-used sections
  • graphify - Turns any input (code, docs, papers, images) into a clustered knowledge graph (HTML + JSON)
  • raginclude-generator - Generates a .raginclude file to curate what a RAG knowledge base should ingest
  • cl-project-init - (ChameleonLabs-specific, sanitized) Example project scaffolder (Next.js SaaS / Python / library templates). Fork and rename for your own company — the cl- prefix marks it as company-scoped.
  • yt-transcript - Download a YouTube transcript into a local yt-transcript project
  • product-manager - Full PM toolkit (strategy, discovery, market research, GTM, execution)
  • install-claude-config - Merges this repo's config into ~/.claude/ with diff-and-ask conflict resolution
  • sync-claude-config - Syncs ~/.claude/ back to this repo with automatic sanitization of private info
  • time - Tracks execution time for each step in a workflow

CodeMap System

The CodeMap is a structured documentation file (Docs/CodeMap.md) that provides:

  • Line-number-accurate class/method inventory
  • Visual ASCII architecture diagrams
  • Cross-file dependency mapping
  • Multi-level tables of contents
  • Language-specific guidelines (Python, C#, JavaScript/TypeScript, XAML)

Instructions Directory

On-demand reference files that Claude Code reads only when needed (reduces context bloat):

  • credentials.md - Platform-specific secret storage patterns
  • file-operations.md - Absolute path conventions and parallel search patterns
  • plan-templates.md - Implementation checklist format with status markers

Output Styles

Three output styles for different contexts:

  • GenUI - Generates self-contained HTML documents with modern styling
  • HTML - Web development focus with accessibility and standards compliance
  • Technical Quality - Comprehensive analysis with systematic problem-solving

Status Line

statusline-command.sh renders a two-line status bar: user@host, working directory, and git branch on line 1; model, context usage, PR badge (color-coded by review state), and vim mode on line 2. When the current session has a messaging channel attached (Discord, Telegram), the channel name appears in its brand color with an "active" suffix — detected live via the process tree, not just pairing files.

Design Philosophy

Context Efficiency

The CLAUDE.md is kept under ~200 lines by extracting detailed reference material to instructions/ files. This reduces /resume time and token usage while keeping critical rules always loaded.

Security First

  • Credentials are never stored in project directories
  • Bash commands never contain inline secrets
  • All issues/inputs are treated as untrusted (prompt injection awareness)
  • Supply-chain defense: 7-day minimum package age enforced across npm/pnpm/pip/uv/poetry/cargo (see CLAUDE.md)
  • GitHub Actions: pull_request_target is banned — the #1 source of Actions supply-chain compromises

Convention Over Configuration

  • Projects follow their own naming conventions (checked before creating files)
  • CodeMaps use standardized format across all languages
  • Plan files use consistent checklist markers

Authors

Created by Leland Green (@lelandg) and Claude (Opus 4.6) at Chameleon Labs.

Join our Discord server - free AI chat with the latest Claude, Gemini, and ChatGPT pro models.

Contributing

Found something useful to add? PRs welcome. Keep these principles in mind:

  • No private information (names, paths, credentials, even as examples)
  • Keep CLAUDE.md concise - extract detail to instructions/
  • Agents should be generic enough for any project
  • Skills should be self-contained with reference files

License

MIT - Use however you like.

Ver o original no GitHub

Bom para

Relacionados