Claude Code Configuration Repository

A comprehensive, ready-to-use .claude configuration repository covering Skills, Subagents, MCP Servers, Hooks, Keybindings, and Settings everything you need to supercharge Claude Code.

Repository Structure

.
├── README.md
├── .gitignore
├── .mcp.json                          # MCP server configurations (project-level)
│
├── .claude/
│   ├── CLAUDE.md                      # Project instructions for Claude
│   ├── settings.json                  # Project-level settings
│   ├── settings.local.json            # Local overrides (gitignored)
│   │
│   ├── skills/                        # Custom slash commands
│   │   ├── commit/SKILL.md            # /commit — smart git commits
│   │   ├── review-pr/SKILL.md         # /review-pr — PR review
│   │   ├── test/SKILL.md              # /test — run & fix tests
│   │   ├── security-audit/SKILL.md    # /security-audit — security scan
│   │   ├── generate-docs/SKILL.md     # /generate-docs — write docs
│   │   └── deploy/SKILL.md            # /deploy — deployment helper
│   │
│   └── agents/                        # Custom subagents
│       ├── code-reviewer.md           # Deep code review specialist
│       ├── test-generator.md          # Test writing specialist
│       ├── security-auditor.md        # Security analysis specialist
│       └── doc-writer.md              # Documentation specialist
│
├── hooks/                             # Hook scripts
│   ├── pre-tool-use.sh
│   ├── post-tool-use.sh
│   └── session-start.sh
│
└── examples/
    ├── user-settings.json             # ~/.claude/settings.json example
    ├── keybindings.json               # ~/.claude/keybindings.json example
    └── managed-mcp.json               # Enterprise managed MCP example

Quick Start

1. Clone this repo into your project

git clone https://github.com/your-username/claude-config .claude-config
cp -r .claude-config/.claude .
cp .claude-config/.mcp.json .

2. Or use as a template

Click "Use this template" on GitHub to create your own copy.

3. Install for a single project

# Copy settings
cp examples/user-settings.json ~/.claude/settings.json

# Copy keybindings
cp examples/keybindings.json ~/.claude/keybindings.json

Settings

Configuration Scopes (Priority Order)

Scope	File	Shared	Priority
Managed	System dirs (see below)	Yes	Highest
User	`~/.claude/settings.json`	No	2nd
Project	`.claude/settings.json`	Yes (git)	3rd
Local	`.claude/settings.local.json`	No	Lowest

Managed settings file locations:

macOS: /Library/Application Support/ClaudeCode/managed-settings.json
Linux: /etc/claude-code/managed-settings.json
Windows: C:\Program Files\ClaudeCode\managed-settings.json

Key Settings Options

{
  // API Configuration
  "apiBaseUrl": "https://api.anthropic.com",
  "model": "claude-opus-4-6",

  // Permission mode: "default" | "acceptEdits" | "autoEdit" | "bypassPermissions"
  "permissionMode": "default",

  // Tool permissions
  "allowedTools": ["Read", "Glob", "Grep"],
  "deniedTools": ["Bash"],

  // Disable non-essential features for performance
  "disableAutoUpdates": false,
  "disableTelemetry": false,

  // Environment variables passed to Claude
  "env": {
    "ANTHROPIC_API_KEY": "${ANTHROPIC_API_KEY}",
    "MAX_MCP_OUTPUT_TOKENS": "25000"
  },

  // Hooks configuration (see Hooks section)
  "hooks": {},

  // MCP tool search (auto-enabled when tools > 10% context)
  "enableMcpToolSearch": true
}

See .claude/settings.json for the full working example.

CLAUDE.md

CLAUDE.md is loaded at session start and provides persistent instructions to Claude. Place it at:

~/.claude/CLAUDE.md — applies to all projects (user-level)
.claude/CLAUDE.md — applies to this project only
Any parent directory — applies to all subdirectories

What to put in CLAUDE.md

# Project Instructions

## Tech Stack
- Language: TypeScript / Node.js
- Framework: Next.js 14
- Database: PostgreSQL with Prisma ORM
- Testing: Jest + Playwright

## Code Conventions
- Use `const` over `let` where possible
- Always add JSDoc to exported functions
- Run `npm test` before committing

## Important Rules
- Never modify files in `generated/` directly
- Always use environment variables for secrets
- Prefer functional components over class components

See .claude/CLAUDE.md for the full example.

Skills (Slash Commands)

Skills live in .claude/skills/<name>/SKILL.md. Invoke with /<name> in Claude Code.

SKILL.md Frontmatter Reference

---
name: skill-name              # Lowercase, hyphens, max 64 chars
description: |                # Tells Claude WHEN to auto-invoke
  When to use this skill...
argument-hint: "[optional]"   # Shown in autocomplete
disable-model-invocation: false   # true = only user can invoke
user-invocable: true          # false = only Claude can invoke
allowed-tools: "Read, Grep"   # Tools usable without permission
model: claude-haiku-4-5-20251001  # Override model for this skill
context: fork                 # "fork" = run in isolated subagent
agent: Explore                # Built-in agent type to use
---

Skill prompt content here...

String Substitutions

Variable	Description
`$ARGUMENTS`	All arguments passed to the skill
`$ARGUMENTS[0]`, `$0`	First argument
`$ARGUMENTS[1]`, `$1`	Second argument
`${CLAUDE_SESSION_ID}`	Current session ID
`${CLAUDE_SKILL_DIR}`	Path to this skill's directory

Dynamic Context (Backtick Injection)

Run shell commands whose output is injected before the prompt:

!`git log --oneline -10`

Based on recent commits above, write a changelog entry.

Included Skills

Skill	Command	Description
Smart Commit	`/commit`	Stages, reviews, and commits with conventional message
PR Review	`/review-pr [number]`	Full pull request code review
Test Runner	`/test [file]`	Run tests and auto-fix failures
Security Audit	`/security-audit`	OWASP-based security scan
Generate Docs	`/generate-docs [path]`	Create/update documentation
Deploy	`/deploy [env]`	Deployment checklist and execution

Subagents

Custom subagents live in .claude/agents/<name>.md. Claude automatically delegates tasks to them based on their description.

Agent Frontmatter Reference

---
name: agent-name              # Identifier used in Agent tool calls
description: |                # CRITICAL: tells Claude when to delegate
  Use this agent when...      # Be specific about trigger conditions
tools:                        # Comma-separated or array of allowed tools
  - Read
  - Grep
  - Glob
  - Bash
model: claude-haiku-4-5-20251001  # Optional: use cheaper model
---

System prompt for the agent...

Built-in Subagents

Agent	Description
`general-purpose`	Default agent for complex multi-step tasks
`Explore`	Fast codebase exploration (search, read, grep)
`Plan`	Architecture and implementation planning

Included Custom Agents

Agent	File	Specialization
Code Reviewer	`agents/code-reviewer.md`	Deep code quality analysis
Test Generator	`agents/test-generator.md`	Writing comprehensive tests
Security Auditor	`agents/security-auditor.md`	Security vulnerability scanning
Doc Writer	`agents/doc-writer.md`	Technical documentation

MCP Servers

MCP (Model Context Protocol) connects Claude to external tools. Configured in .mcp.json (project-level, shared via git).

Add a Server

# HTTP server
claude mcp add --transport http <name> <url>

# Stdio server (local process)
claude mcp add --transport stdio <name> -- <command> [args]

# With environment variables
claude mcp add --transport stdio github -- \
  -e GITHUB_TOKEN=$GITHUB_TOKEN \
  npx -y @modelcontextprotocol/server-github

MCP Scopes

Scope	Storage	Shared	Command
Local	`~/.claude.json`	No	`--scope local` (default)
Project	`.mcp.json`	Yes (git)	`--scope project`
User	`~/.claude.json`	No	`--scope user`

Precedence: Local > Project > User

Included MCP Servers (`.mcp.json`)

Server	Transport	Use Case
GitHub	Stdio	Issues, PRs, repos
Filesystem	Stdio	File operations
Memory	Stdio	Persistent knowledge graph
Brave Search	HTTP	Web search
PostgreSQL	Stdio	Database queries
Slack	HTTP	Team messaging
Notion	HTTP	Documentation

MCP Commands

claude mcp list                    # List all servers
claude mcp get <name>              # Server details
claude mcp remove <name>           # Remove server
claude mcp reset-project-choices   # Reset approval state
/mcp                               # Check status in Claude Code

Hooks

Hooks run shell commands, HTTP requests, prompts, or agents at lifecycle events.

Hook Events

Event	Trigger
`SessionStart`	Session begins or resumes
`UserPromptSubmit`	Before Claude processes your message
`PreToolUse`	Before any tool executes
`PostToolUse`	After tool succeeds
`PostToolUseFailure`	After tool fails
`PermissionRequest`	Permission dialog triggered
`Stop`	Claude finishes responding
`SubagentStart`	Subagent spawned
`SubagentStop`	Subagent finishes
`Notification`	Notification sent
`TaskCompleted`	Task marked complete
`PreCompact`	Before context compaction
`PostCompact`	After context compaction
`SessionEnd`	Session terminates

Hook Configuration

{
  "hooks": {
    "PreToolUse": [
      {
        "matcher": "Bash",
        "hooks": [
          {
            "type": "command",
            "command": "~/.claude/hooks/pre-tool-use.sh",
            "timeout": 10
          }
        ]
      }
    ],
    "PostToolUse": [
      {
        "matcher": ".*",
        "hooks": [
          {
            "type": "http",
            "url": "http://localhost:9000/hook",
            "headers": { "Authorization": "Bearer $HOOK_TOKEN" }
          }
        ]
      }
    ]
  }
}

Hook Types

Command Hook:

{
  "type": "command",
  "command": "/path/to/script.sh",
  "timeout": 600,
  "async": false,
  "statusMessage": "Validating..."
}

HTTP Hook:

{
  "type": "http",
  "url": "http://localhost:8080/hooks/event",
  "headers": { "X-Secret": "$HOOK_SECRET" },
  "allowedEnvVars": ["HOOK_SECRET"],
  "timeout": 30
}

Prompt Hook:

{
  "type": "prompt",
  "prompt": "Review this tool call: $ARGUMENTS",
  "model": "claude-haiku-4-5-20251001",
  "timeout": 30
}

Agent Hook:

{
  "type": "agent",
  "prompt": "Verify tests pass after this change.",
  "timeout": 60
}

Hook Exit Codes (Command Hooks)

Code	Meaning
`0`	Success; stdout parsed as JSON response
`2`	Blocking error; action is blocked
Other	Non-blocking warning; shown in verbose mode

MCP Tool Matcher Pattern

mcp__<server>__<tool>

Examples:
  mcp__github__.*          # All GitHub tools
  mcp__.*__write.*         # All write tools across servers
  mcp__memory__.*          # All Memory tools

See .claude/settings.json and hooks/ for working examples.

Keybindings

Keybindings live at ~/.claude/keybindings.json (user-level, not project-level).

File Format

{
  "$schema": "https://www.schemastore.org/claude-code-keybindings.json",
  "$docs": "https://code.claude.com/docs/en/keybindings",
  "bindings": [
    {
      "context": "Chat",
      "bindings": {
        "ctrl+e": "chat:externalEditor",
        "ctrl+u": null
      }
    }
  ]
}

Available Contexts

Global, Chat, Autocomplete, Settings, Confirmation, Tabs, Help, Transcript, HistorySearch, Task, ThemePicker, Attachments, Footer, MessageSelector, DiffDialog, ModelPicker, Select, Plugin

Key Syntax

ctrl+k           # Single modifier
alt+shift+p      # Multiple modifiers
ctrl+k ctrl+s    # Chord (sequence)
escape, enter, tab, space, up, down, left, right
backspace, delete, f1–f12

Common Actions

Context	Action	Default
Global	`app:interrupt`	`ctrl+c`
Global	`app:exit`	`ctrl+d`
Global	`app:toggleTodos`	`ctrl+t`
Chat	`chat:submit`	`enter`
Chat	`chat:cancel`	`escape`
Chat	`chat:externalEditor`	`ctrl+g`
Chat	`chat:stash`	`ctrl+s`
Autocomplete	`autocomplete:accept`	`tab`
Confirmation	`confirm:yes`	`y` / `enter`
Confirmation	`confirm:no`	`n` / `escape`

Set null to unbind a key: "ctrl+u": null

See examples/keybindings.json for a full working example.

Memory System

Claude Code has a built-in memory system at ~/.claude/projects/<project>/memory/.

Memory Types

Type	When to Use
`user`	User role, preferences, expertise level
`feedback`	Corrections and behavioral guidance
`project`	Goals, decisions, deadlines
`reference`	External system pointers (Jira, Slack, etc.)

Memory File Format

---
name: memory-name
description: One-line description for relevance matching
type: user | feedback | project | reference
---

Memory content here.

**Why:** Reason this memory exists.
**How to apply:** When/where to use this.

Configuration Scopes

Priority (highest → lowest):

[Managed]    /etc/claude-code/managed-settings.json
             C:\Program Files\ClaudeCode\managed-settings.json  (Windows)
             /Library/Application Support/ClaudeCode/           (macOS)
     ↓
[User]       ~/.claude/settings.json
     ↓
[Project]    .claude/settings.json
     ↓
[Local]      .claude/settings.local.json   ← never commit this

Gitignore Recommendations

# Always gitignore
.claude/settings.local.json
.env
.env.local

# Optional: gitignore if contains secrets
.mcp.json

Resources

Contributing

Contributions are welcome! This repository grows better with community-shared skills, agents, MCP server configs, hooks, and keybinding setups.

What you can contribute

Type	Where to add	What makes a good one
Skill	`.claude/skills/<name>/SKILL.md`	Clear `description` so Claude knows when to auto-invoke; includes examples
Subagent	`.claude/agents/<name>.md`	Focused on a specific domain; specifies appropriate `tools` and `model`
MCP Server	`.mcp.json`	Commented with setup instructions, required env vars, and a link to the server
Hook	`hooks/<event-name>.sh`	Well-commented, safe exit codes, handles all three OSes
Keybinding preset	`examples/keybindings/`	Named for the use-case (e.g., `vim-style.json`, `minimal.json`)
Settings preset	`examples/settings/`	Named for the context (e.g., `enterprise.json`, `open-source.json`)

How to contribute

Fork this repository
Create a branch: git checkout -b feat/my-skill-name
Add your file(s) following the existing structure and conventions
Test it — make sure your skill/agent works in Claude Code
Open a Pull Request with a short description of what it does and when to use it

Guidelines

Keep skill and agent description fields clear and specific — Claude uses them to decide when to delegate
Never include real API keys, tokens, or secrets — use ${ENV_VAR} placeholders
Add comments explaining non-obvious configuration choices
One skill or agent per PR keeps reviews focused

About Author

Name: Muhammad Imran Zaman

Email: imranzaman.ml@gmail.com

Professional Links:

Name		Name	Last commit message	Last commit date
Latest commit History 2 Commits
.claude		.claude
examples		examples
hooks		hooks
.gitattributes		.gitattributes
.gitignore		.gitignore
.mcp.json		.mcp.json
README.md		README.md

Folders and files

Latest commit

History

Repository files navigation

Claude Code Configuration Repository

Table of Contents

Repository Structure

Quick Start

1. Clone this repo into your project

2. Or use as a template

3. Install for a single project

Settings

Configuration Scopes (Priority Order)

Key Settings Options

CLAUDE.md

What to put in CLAUDE.md

Skills (Slash Commands)

SKILL.md Frontmatter Reference

String Substitutions

Dynamic Context (Backtick Injection)

Included Skills

Subagents

Agent Frontmatter Reference

Built-in Subagents

Included Custom Agents

MCP Servers

Add a Server

MCP Scopes

Included MCP Servers (.mcp.json)

MCP Commands

Hooks

Hook Events

Hook Configuration

Hook Types

Hook Exit Codes (Command Hooks)

MCP Tool Matcher Pattern

Keybindings

File Format

Available Contexts

Key Syntax

Common Actions

Memory System

Memory Types

Memory File Format

Configuration Scopes

Gitignore Recommendations

Resources

Contributing

What you can contribute

How to contribute

Guidelines

About Author

About

Topics

Resources

Uh oh!

Stars

Watchers

Forks

Releases 1

Packages 0

Uh oh!

Contributors

Uh oh!

Languages

Included MCP Servers (`.mcp.json`)

Packages