Claude Code Complete Guide

This guide covers every discoverable Claude Code command as of August 2025,
including many features that are not widely known or documented in basic tutorials.

This represents the most complete Claude Code command reference available.

---

For updates and contributions, visit the official Claude Code documentation

Section	Status
Guides on how to install on Windows, Linux, MacOS	✅
Tips and Tricks	✅
MCP Overview with what to use	✅
Community Guides	✅
Troubleshooting	✅
How to use Claude code the most optimal way	✅

I Usually Start my Claude with Claude --dangerously-skip-permissions only use this if you know exactly what ur doing!

Also use /agents to create/edit subagents that can be called with @agent-name when using claude code or simply by refering to it.

---

Table of Contents

1. Quick Start
2. System requirements
3. Health Check & Error Fixes
4. Sub Agents
5. Claude Commands
6. MCP Integration
7. Configuration
8. Environment Variables
9. Security & Permissions
10. Claude CLI Configuration
11. Claude ~/.claude.json Configuration Guide
12. Automation & Scripting
13. Troubleshooting
14. Advanced Features
15. Best Practices

---

Quick Start

## NPM (global) ⭐️ Official
npm install -g @anthropic-ai/claude-code
# if only typing "claude" does not work try "npx claude"

## Windows 
npm install -g @anthropic-ai/claude-code
irm https://claude.ai/install.ps1 | iex

## WSL/GIT 
npm install -g @anthropic-ai/claude-code
curl -fsSL https://claude.ai/install.sh | bash

## MacOS
brew install node
npm install -g @anthropic-ai/claude-code
# issue with claude bot found? run export PATH="$PATH:$(npm bin -g)"

## Linux:
sudo apt update && sudo apt install -y nodejs npm
npm install -g @anthropic-ai/claude-code

curl -fsSL https://claude.ai/install.sh | bash

## Arch Linux AUR
yay -S claude-code 

## Docker (containerised)
mkdir -p ~/.docker/cli-plugins
curl -SL https://github.com/docker/buildx/releases/download/v0.11.2/buildx-v0.11.2.linux-amd64 -o ~/.docker/cli-plugins/docker-buildx
chmod +x ~/.docker/cli-plugins/docker-buildx
curl -O https://raw.githubusercontent.com/RchGrav/claudebox/main/claudebox
chmod +x claudebox
# Nice when you can't touch the host system

# You can also open claude code with ur entire project in vscode or cursor by following this:
# cd in the directory u want to work with then run the command
code .
# Have claude code extention installed!
# This also works if your using WSL

## check if claude is installed
which claude
claude --version


# Interactive Mode
claude                      # Start interactive REPL
claude "your question"      # Start with initial prompt

# One-Shot Mode  
claude -p "analyze this"    # Quick query and exit
cat file | claude -p "fix"  # Process piped content

# Management
claude config              # Configure settings
claude update              # Update to latest
claude mcp                 # Setup MCP servers
claude /agents             # Configure/Setup Subagents for different tasks 

System requirements

OS: macOS 10.15+, Ubuntu 20.04+/Debian 10+, or Windows 10/11 or WSL
Hardware: 4GB RAM minimum 8GB+ recommended
Software: Node.js 18+ or git 2.23+ (optional) & GitHub or GitLab CLI for PR workflows (optional)
Internett: Connection for API calls

Initial Setup

1. API Key Configuration

# Required: Get your API key from https://console.anthropic.com
export ANTHROPIC_API_KEY="sk-your-key-here"

# Make permanent (choose your shell)
# Bash
echo 'export ANTHROPIC_API_KEY="sk-your-key-here"' >> ~/.bashrc
source ~/.bashrc

# Zsh
echo 'export ANTHROPIC_API_KEY="sk-your-key-here"' >> ~/.zshrc
source ~/.zshrc

# Fish
echo 'set -gx ANTHROPIC_API_KEY "sk-your-key-here"' >> ~/.config/fish/config.fish

2. Basic Configuration

# Interactive setup
claude config

# Set basic defaults 
claude config set -g verbose true
claude config set -g outputFormat text

# Test installation
claude "Hello, Claude!"
claude /doctor
claude update

3. Settings i Turn Off

export DISABLE_TELEMETRY=1
export DISABLE_ERROR_REPORTING=1
export DISABLE_NON_ESSENTIAL_MODEL_CALLS=1


# Security defaults
claude config set allowedTools "Edit,View"
claude config set hasTrustDialogAccepted
claude config set hasCompletedProjectOnboarding 
claude config set ignorePatterns
claude config set --global

Health Check

claude                         # opens Claude UI (if on PATH)
where claude                   # shows path(s), e.g. C:\Users\<you>\AppData\Roaming\npm\claude.cmd
claude /doctor                 # opens diagnostic/debug window
"%USERPROFILE%\AppData\Roaming\npm\claude.cmd" --version   # e.g. 1.0.85 (Claude Code)
--------------------------------------------
# CANT RUN 'claude' BUT 'npx claude' WORKS? In most cases it's PATH that's broken/missing. try a quick temporary fix (current CMD only):
set PATH=%USERPROFILE%\AppData\Roaming\npm;C:\Program Files\nodejs;%PATH%
where claude
claude doctor
--------------------------------------------
# Permanent, safe fix (Windows GUI):
1. Start → type "Environment Variables" → Open "Edit the system environment variables" → Environment Variables…
2. Under "User variables for <you>" select Path → Edit → Add:
   C:\Users\<you>\AppData\Roaming\npm
   (optional: C:\Users\<you>\.claude\local\bin and C:\Users\<you>\.local\bin)
3. Remove duplicates, any entry containing %PATH% or stray quotes (") and click OK.
4. Open a NEW Command Prompt and verify:
   where claude
   claude doctor
--------------------------------------------
# If still failing, run the shim directly:
npx claude doctor
# or
"%USERPROFILE%\AppData\Roaming\npm\claude.cmd" doctor
--------------------------------------------
# If none of these worked check if "npm" works in ur terminal if it does "npx claude" will also work then you know nothing is wrong with your node.js path try reinstalling claude code
npm uninstall -g @anthropic-ai/claude-code
npm install -g @anthropic-ai/claude-code
--------------------------------------------
# If this did not work try completely reinstalling
npm uninstall -g @anthropic-ai/claude-code
# Remove any leftover shim files (delete if they exist)
Remove-Item -LiteralPath "$env:USERPROFILE\AppData\Roaming\npm\claude*" -Force -ErrorAction SilentlyContinue
Remove-Item -LiteralPath "$env:USERPROFILE\AppData\Roaming\npm\node_modules\@anthropic-ai\claude-code" -Recurse -Force -ErrorAction SilentlyContinue

# Delete cached installer & installed native files
Remove-Item -LiteralPath "$env:USERPROFILE\.claude\downloads\*" -Recurse -Force -ErrorAction SilentlyContinue
Remove-Item -LiteralPath "$env:USERPROFILE\.claude\local\bin\claude.exe" -Force -ErrorAction SilentlyContinue
Remove-Item -LiteralPath "$env:USERPROFILE\.claude\local" -Recurse -Force -ErrorAction SilentlyContinue

# Remove config and project-local files
Remove-Item -LiteralPath "$env:USERPROFILE\.claude.json" -Force -ErrorAction SilentlyContinue
Remove-Item -LiteralPath "$env:USERPROFILE\.claude" -Recurse -Force -ErrorAction SilentlyContinue

then install and try "claude" again
npm i -g @anthropic-ai\claude-code

---

Claude Commands

Claude / Commands

Slash Cmd	Purpose
/help	List slash commands
/agents	List/Create/Edit Subagents
/add-dir	Add more working dirs
/bug	Report bug to Anthropic
/clear	Clear chat history
/compact	Compact conversation
/config	Config menu
/cost	Token usage
/doctor	Health check
/exit	Exit REPL
/init	Generate CLAUDE.md
/login / /logout	Auth switch
/mcp	Manage MCP servers
/memory	Edit memories
/model	Change model
/permissions	Tool permissions
/pr_comments	View PR comments
/review	Request code review
/sessions	List sessions
/status	System/account status
/terminal-setup	Install Shift+Enter binding
/vim	Toggle vim mode
/output-style	Show available output styles with descriptions

Claude -- Commands

Session-related flags (--continue, --resume) are listed in Session Commands

Flag	Short	Options / Value	Description	Example
--print	-p		Non-interactive “print” mode	claude -p "help"
--help	-h		Show help	claude --help
--version	-v		Show version	claude --version
--model <name>			Choose model	claude --model claude-opus-4-20250514
--verbose			Verbose logs	claude --verbose
--add-dir <paths…>			Add extra working directories	claude --add-dir ../lib ../src
--mcp-config <file>			Load MCP server config	claude --mcp-config servers.json
--output-format		text, json, stream-json	Output format (works great with -p)	--output-format json
--input-format		text, stream-json	Input format for -p	--input-format stream-json
--max-turns <n>		Integer	Limit number of turns with -p	--max-turns 3
--system-prompt "<text>"		String	Set system prompt in -p	--system-prompt "You are strict"
--append-system-prompt "<text>"		String	Append to system prompt in -p	--append-system-prompt "Add tests"
--allowedTools "<list>"		Comma/space list	Whitelist tools	--allowedTools "Read,View"
--disallowedTools "<list>"		Comma/space list	Blacklist tools	--disallowedTools "Bash"
--permission-mode <mode>		e.g., plan	Start in permission mode	--permission-mode plan
--permission-prompt-tool <tool>		Tool name	MCP tool for permission checks	--permission-prompt-tool mcp__auth__prompt
--dangerously-skip-permissions			Skip all permission prompts (⚠️)	--dangerously-skip-permissions

Claude Session Commands

Command / Flag	What it does	Example
claude -c / claude --continue	Continue last session	claude -c
claude -c -p "<prompt>"	Continue + new prompt in print	claude -c -p "check types"
claude -r <id> / claude --resume <id|name>	Resume by session ID or saved name	claude --resume sprint-review

Tip: For non-interactive runs, you can also pipe input: cat logs.txt \| claude -p "explain"

Subcommands:

Subcommand	Description	Example
claude update	Self-update to latest version	claude update
claude mcp	Launch MCP wizard	claude mcp
claude config	Interactive config wizard	claude config
claude config list	List all keys	claude config list
claude config get <key>	Get value	claude config get theme
claude config set <key> <val>	Set value	claude config set theme dark
claude config add <key> <vals…>	Append to array	claude config add env DEV=1
claude config remove <key> <vals…>	Remove items	claude config remove env DEV=1

---

Sub Agents

Sub‑Agents are purpose‑built helpers with their own prompts, tools, and isolated context windows. Treat this like a “mixture‑of‑experts” you compose per repo.

When to use them

• You need high signal responses (plans, reviews, diffs) without side quests.
• You want version‑controlled prompts and tool policies alongside the codebase.
• You work in PR‑driven teams and want scoped edits by role.

Each Sub‑Agent Has Its Own Context

Design rules for your lineup

• Define one clear responsibility per agent.
• Keep the minimum tool set needed for that role.
• Prefer read‑only agents for analysis/review tasks.
• Give edit powers to as few agents as possible.

Caption: Agents selection UI in the terminal.

How I Configure Agents

Keep agents in the project so they’re versioned with the repo and evolve via PRs.

Quick start

# Update CLI and open the agents panel
claude update
/agents

Create your core agents

• planner (read‑only): turns features/issues into small, testable tasks; outputs a task list or plan.md.
• codegen (edit‑capable): implements tasks; limited to src/ + tests/.
• tester (read‑only or patch‑only): writes one failing test or a minimal repro.
• reviewer (read‑only): leaves structured review comments; never edits.
• docs (edit‑capable): updates README.md/docs/ only.

Policy tip: Prefer patch output for edit‑capable agents so changes land through your normal Git workflow.

Caption: Choose only the tools an agent truly needs (e.g., advisory vs editing access).

Example prompts

Keep prompts short, testable, and repo‑specific. Check them into agents/:

Caption: Example prompt for a test‑coverage‑analyzer agent.

tester.prompt.md (sample)

Role: Write a single, focused failing test for the specific scenario I describe.
Scope: Only create/modify tests under tests/. Do not change src/.
Output: A brief rationale + a unified diff or patch.
If the scenario is unclear, ask exactly one clarifying question.

Expected output

Your tester agent should produce a small diff or patch plus a short rationale:

Caption: Example response from the test‑coverage‑analyzer agent.

Acceptance checklist

• Output is a single change set.
• Only files in allowed paths are touched.
• Rationale explains intent and edge cases.
• If blocked, the agent asked one clear question.

---

Why This Shift Matters

Operational benefits

• Less context switching: you stay in one mental mode; agents do the rest.
• Cleaner PRs: narrow prompts + limited tools → smaller, reviewable diffs.
• Fewer regressions: tester/reviewer agents catch gaps before merge.
• Repeatability: prompts + policies live in the repo and travel with branches.

Security & governance

• Limit write access by path (e.g., src/, tests/, docs/).
• Favor read‑only analysis for high‑risk areas.
• Log/commit assistant outputs as patches for auditability.

---

A Mindset Shift

Do

• Treat agents as teammates with job descriptions.
• Start read‑only; grant write access last.
• Keep prompts in version control and iterate via PR.

Don’t

• Ask one agent to plan, code, and test in a single turn.
• Give blanket write permissions.
• Accept multi‑file diffs when you asked for one test.

Cant find Subagents to look at?

• Start with these!
  • Configurations, Agents, Mcps, Templates
  • github-awesome-claude-code-subagents
  • medium-claude-code-subagents-examples-with-templates-you-can-use

🔌 MCP Integration

Understanding MCP (Model Context Protocol)

What is MCP? MCP extends Claude's capabilities by connecting to external services, databases, APIs, and tools.

MCP Architecture:

Claude Code ←→ MCP Protocol ←→ MCP Servers ←→ External Services

MCP Setup & Configuration

Basic MCP Commands

claude mcp                    # Interactive MCP configuration
claude mcp list              # List configured servers            
claude mcp add <name> <cmd>  # Add new server
claude mcp remove <name>     # Remove server

MCP Configuration File

Location:~/.claude.json

Scope-Based Configuration Files

User/Global Scope: Global MCP servers Project Scope: Project-scoped servers are stored in a .mcp.json file at your project's root directory

{
  "mcpServers": {
    "git": {
      "command": "git-mcp-server",
      "args": [],
      "env": {}
    },
    "postgres": {
      "command": "postgres-mcp-server", 
      "args": ["--host", "localhost", "--port", "5432"],
      "env": {
        "POSTGRES_USER": "developer",
        "POSTGRES_PASSWORD": "dev_password",
        "POSTGRES_DB": "myapp_development"
      }
    }
  }
}

MCP Servers

Note: The exact package names and installation commands below may not be accurate. Consult official MCP documentation for current server packages.

Development Tools

# npm install -g git-mcp-server         

# claude mcp add git "git-mcp-server"
# claude mcp add github "github-mcp-server --token $GITHUB_TOKEN"

Database Integration

npm install -g postgres-mcp-server               
npm install -g mysql-mcp-server                  
npm install -g sqlite-mcp-server               

# Setup examples may look like this:
# export POSTGRES_URL="postgresql://user:password@localhost:5432/mydb"
# claude mcp add postgres "postgres-mcp-server --url $POSTGRES_URL"

MCP Tool Permissions

# Allow specific MCP tools 
claude --allowedTools "mcp__git__commit,mcp__git__push"

# Allow all tools from specific server
claude --allowedTools "mcp__postgres__*"

# Combined with built-in tools
claude --allowedTools "Edit,View,mcp__git__*"

---

⚙️ Configuration

System Overview

Claude Code uses a hierarchical configuration system:

1. Command-line flags (highest priority)
2. Environment variables
3. Project configuration (location may vary)
4. Global configuration (likely ~/.claude.json)
5. Built-in defaults (lowest priority)

Configuration Files

Global Configuration

Location: ~/.claude.json

{
  "model": "claude-sonnet-4",
  "verbose": true,
  "outputFormat": "text", 
  "allowedTools": ["Edit", "View"],
  "disallowedTools": [],
}

Project Configuration

Location: settings.json OR similar

{
  "model": "claude-sonnet-4",
  "systemPrompt": "You are a senior developer working on this project",
  "allowedTools": [
    "Edit",
    "View",
    "Bash(git:*)",
    "Bash(npm:*)"
  ],
}

Environment Variables

Core Variables

Variable	Required	Purpose	Example
ANTHROPIC_API_KEY	YES	API Authentication	sk-ant-api03-xxx
ANTHROPIC_MODEL	No	Default model	claude-sonnet-4
ANTHROPIC_BASE_URL	No	API endpoint override	https://api.anthropic.com

Environment Variables

Env Var	Default	Example Value	Effect
DISABLE_NON_ESSENTIAL_MODEL_CALLS	0	1	Skip auto‑summaries, background explanations & git diff scans ⇒ faster, cheaper.
MAX_THINKING_TOKENS	≈30‑40 k	50000	Higher token budget for reading code, analyzing diffs & planning.
DISABLE_TELEMETRY	0	1	Block anonymous usage + error telemetry.
CLAUDE_CODE_USE_BEDROCK	0	1	Route requests via AWS Bedrock (needs IAM creds; falls back if absent).
CLAUDE_CODE_USE_VERTEX	0	1	Route requests via Google Vertex AI (needs service‑account creds; falls back if absent).

Env Var	Default	Example Value	What It Does
HTTP_PROXY	(unset)	http://proxy.company.com:8080	Routes HTTP requests through the given proxy.
HTTPS_PROXY	(unset)	https://proxy.company.com:8443	Routes HTTPS requests through the given proxy.
NO_PROXY	localhost,127.0.0.1	localhost,127.0.0.1,*.company.com	Comma‑separated hosts/IPs that bypass the proxy.

How to Set

Shell (temporary)

export HTTP_PROXY="http://proxy.company.com:8080"
export HTTPS_PROXY="https://proxy.company.com:8443"
export NO_PROXY="localhost,127.0.0.1,*.company.com"
claude "Test request via proxy"

Shell Profile (persistent)

# ~/.bashrc or ~/.zshenv
export HTTP_PROXY="http://proxy.company.com:8080"
export HTTPS_PROXY="https://proxy.company.com:8443"
export NO_PROXY="localhost,127.0.0.1,*.company.com"

Reload with source ~/.bashrc.

GitHub Actions

env:
  HTTP_PROXY:  "http://proxy.company.com:8080"
  HTTPS_PROXY: "https://proxy.company.com:8443"
  NO_PROXY:    "localhost,127.0.0.1,*.company.com"

---

Security & Permissions

Permission System

How it works:

• Claude asks for permission before using tools
• Permissions are remembered per session
• Dangerous operations require confirmation

Permission Levels

Level	Description	Risk	Use Case
Interactive	Prompt for each operation	Low	Development work
Allowlist	Pre-approved tools only	Medium	Automation scripts
Dangerous	Skip all permissions	CRITICAL	Containers only

Tool Permission Patterns

# Allow specific tools choose ur own can also be used like: Bash(*)
claude --allowedTools "Edit,View"

# Allow tool categories  
claude --allowedTools "Edit,View,Bash"

# Scoped permissions (Git operations only)
claude --allowedTools "Bash(git:*)"

# Multiple scopes
claude --allowedTools "Bash(git:*),Bash(npm:*)"

Dangerous Mode (CRITICAL Security Feature)

# DANGEROUS - Can cause data loss
claude --dangerously-skip-permissions

# Only use in isolated environments:
# ✅ Safe: Isolated Docker container  
# ❌ NEVER: Production systems, shared machines, systems with important data

Security Best Practices

1. Start Restrictive

# Good: Specific permissions
claude --allowedTools "Edit,View,Bash(git:status)"

# Bad: Broad permissions  
claude --allowedTools "Bash"

2. Protect Sensitive Data

# Good: Environment variables
export DATABASE_URL="postgresql://user:pass@host/db"

# Bad: Hardcoded credentials in commands
# claude "connect to postgresql://user:password123@host/db"

3. Regular Security Audits

# Check current permissions
claude config get allowedTools
claude config get disallowedTools

# Review configuration
claude config list

---

Thinking Keywords

Claude Code supports extended thinking — extra pre‑answer planning time for harder problems. You can nudge that planning budget with specific trigger words that map to progressively larger “thinking budgets”.

keywords

These map in order of increasing budget: think < think hard < think harder < ultrathink.

Use only the four keywords above. Other phrases (e.g., “think more”, “megathink”, etc.) are not documented and should not be relied on.

What extended thinking does

Before Claude starts producing the final answer, it spends more time:

• planning a solution,
• breaking down steps,
• considering alternatives and trade‑offs,
• checking constraints and edge cases.

How to use it

You can place the keyword anywhere in your prompt (case‑insensitive). If multiple appear, assume the strongest one wins.

# Small boost
claude -p "Think. Outline a plan to refactor the auth module."

# Medium boost
claude -p "Think harder. Draft a migration plan for moving from REST to gRPC."

# Maximum planning budget
claude -p "Ultrathink. Propose a step‑by‑step strategy to fix flaky payments tests and add guardrails."

Notes & caveats

• This is a Claude Code (CLI) behavior, not a public API parameter; naming or effects may evolve over time.
• Higher budgets usually increase latency and token usage. Prefer the smallest keyword that gets the job done.
• Keep prompts crisp. The keyword asks Claude to plan; your prompt should still provide goals, constraints, and success criteria.

Claude CLI Configuration

Configuration keys

---

Prerequisites

1. Authenticate first

   # Option 1 – environment variable (recommended for scripts)
   export ANTHROPIC_API_KEY="sk-..."
   
   # Option 2 – interactive login inside Claude REPL
   claude /login

2. Back‑up current config

   cp ~/.claude/claude.json ~/.claude/claude.json.bak
   
   # This depends where your .json is installed it may also be at ~/.claude/local/package.json

If apiKeyHelper is mis‑configured or no API key is found, you'll see errors like:

Error getting API key from apiKeyHelper (in settings or ~/.claude.json):

Fix authentication before modifying other keys. (docs.anthropic.com)

---

claude config Commands

Command pattern	Purpose	Example
claude config list	Show all current settings	claude config list
claude config get <key>	Display a single setting	claude config get theme
claude config set -g <key> <value>	Set a global value	claude config set -g theme dark
claude config add -g <key> <value>	Append to an array‑type setting	claude config add -g env CLAUDE_CODE_ENABLE_TELEMETRY=1
claude config remove -g <key> <value>	Remove from an array‑type setting	claude config remove -g env CLAUDE_CODE_ENABLE_TELEMETRY

(omit -g to target the current project instead of global).

---

Editable Keys

Key	Typical Values	Safe Example	Notes
apiKeyHelper	Path to executable script	claude config set -g apiKeyHelper ~/.claude/key_helper.sh	Script must echo a fresh API key; be executable. (docs.anthropic.com)
installMethod	npm, brew, binary, deb, …	claude config set -g installMethod npm	Informational only. (ainativedev.io)
autoUpdates	true / false	claude config set -g autoUpdates false	Turns self‑updater on/off. (docs.anthropic.com)
theme	dark, light, light-daltonized, dark-daltonized	claude config set -g theme dark	CLI colour scheme. (docs.anthropic.com)
verbose	true / false	claude config set -g verbose true	Show full Bash + tool output. (docs.anthropic.com)
		`claude config set -g	Where alerts appear. (docs.anthropic.com)
shiftEnterKeyBindingInstalled	true / false	claude config set -g shiftEnterKeyBindingInstalled true	Enables Shift+Enter new‑line. (ainativedev.io)
hasUsedBackslashReturn	true / false	claude config set -g hasUsedBackslashReturn true	Internal flag; rarely changed. (ainativedev.io)
supervisorMode	true / false	claude config set -g supervisorMode true	Enables supervisor features. (ainativedev.io)
autoCompactEnabled	true / false	claude config set -g autoCompactEnabled true	Auto‑compresses chat logs. (ainativedev.io)
diffTool	Diff command/path	claude config set -g diffTool meld	Used by /diff. (ainativedev.io)
env	KEY=value or JSON	claude config set -g env CLAUDE_CODE_ENABLE_TELEMETRY=0	Injects env vars into every session. (docs.anthropic.com)
tipsHistory	[] or JSON array	claude config set -g tipsHistory []	Clears tips pop‑up history. (ainativedev.io)
parallelTasksCount	Integer ≥ 1	claude config set -g parallelTasksCount 4	Limit concurrent tasks. (ainativedev.io)
todoFeatureEnabled	true / false	claude config set -g todoFeatureEnabled true	Enables experimental To‑Do. (ainativedev.io)
messageIdleNotifThresholdMs	Integer (ms)	claude config set -g messageIdleNotifThresholdMs 60000	Idle threshold before alert. (ainativedev.io)
autoConnectIde	true / false	claude config set -g autoConnectIde true	Auto‑connects to IDE at launch. (ainativedev.io)

🔒 Attempting to set any other key (e.g. model) will throw Error: Cannot set '<key>'. Only these keys can be modified… for more check (docs.anthropic.com)

---

Migrating to settings.json

Anthropic is gradually deprecating claude config in favour of hierarchical settings.json files:

# Global (user‑level) settings
vi ~/.claude/settings.json

# Project‑level (checked into git)
vi .claude/settings.json

---

Safe Editing Checklist

1. Backup ~/.claude/claude.json.
2. Authenticate (ANTHROPIC_API_KEY or /login).
3. Change one key at a time → verify with claude config get.
4. Keep CLI updated (autoUpdates=true) or via package manager.
5. Read release notes for new or removed keys.

---

Claude ~/.claude.json Configuration Guide

Purpose — A concise, fact‑checked reference for safely editing your personal configuration file. All keys and examples come directly from Anthropic‑supplied defaults or the CLI's own output—no speculative or undocumented fields.

---

1 ▸ Back Up First

cp ~/.claude.json ~/.claude.json.backup

If anything breaks, restore with:

cp ~/.claude.json.backup ~/.claude.json

---

2 ▸ MCP Servers

mcpServers lets Claude Code interact with external tools (filesystem, web, GitHub, …). Each entry follows the exact schema below.

{
  "mcpServers": {
    "server-name": {
      "type": "stdio",
      "command": "npx",
      "args": ["-y", "package-name"],
      "env": {}
    }
  }
}

2.1 Schema

Field	Required?	Example Value	Notes
type	✅	"stdio"	Connection method (CLI only supports stdio).
command	✅	"npx"	Executable run by Claude Code.
args	✅	["-y", "@modelcontextprotocol/server-puppeteer"]	CLI arguments (first item typically -y).
env	✅	{ "API_KEY": "value" }	Key‑value pairs exported to the child process.

2.2 Ready‑to‑Copy Examples

{
  "mcpServers": {
    "sequential-thinking": {
      "type": "stdio",
      "command": "npx",
      "args": ["-y", "@modelcontextprotocol/server-sequential-thinking"],
      "env": {}
    },
    "puppeteer": {
      "type": "stdio",
      "command": "npx",
      "args": ["-y", "@modelcontextprotocol/server-puppeteer"],
      "env": {}
    },
    "fetch": {
      "type": "stdio",
      "command": "npx",
      "args": ["-y", "@kazuph/mcp-fetch"],
      "env": {}
    }
  }
}

With API keys

{
  "mcpServers": {
    "github": {
      "type": "stdio",
      "command": "npx",
      "args": ["-y", "@modelcontextprotocol/server-github"],
      "env": { "GITHUB_TOKEN": "<your‑token>" }
    },
    "brave-search": {
      "type": "stdio",
      "command": "npx",
      "args": ["-y", "@modelcontextprotocol/server-brave-search"],
      "env": { "BRAVE_API_KEY": "<your‑key>" }
    }
  }
}

More popular servers

{
  "mcpServers": {
    "filesystem": {
      "type": "stdio",
      "command": "npx",
      "args": ["-y", "@modelcontextprotocol/server-filesystem", "/path/to/allowed/directory"],
      "env": {}
    },
    "context7": {
      "type": "stdio",
      "command": "npx",
      "args": ["-y", "@upstash/context7-mcp"],
      "env": {}
    }
  }
}

---

3 ▸ Feature Flags

All three flags below are safe to toggle. Booleans only.

Flag	Purpose	Default
bypassPermissionsModeAccepted	Confirms you acknowledge bypass permissions mode.	false
hasAcknowledgedCostThreshold	Suppresses cost pop‑ups after first confirmation.	false
isQualifiedForDataSharing	Opt‑in/out of anonymous telemetry.	false

Example:

{
  "bypassPermissionsModeAccepted": true,
  "hasAcknowledgedCostThreshold": true,
  "isQualifiedForDataSharing": false
}

---

4 ▸ Reset Tips & Onboarding

{
  "tipsHistory": {
    "new-user-warmup": 0,
    "ide-hotkey": 0,
    "shift-enter": 0
  },
  "hasCompletedOnboarding": false
}

Set counters to 0 or hasCompletedOnboarding to false to see onboarding screens again.

---

5 ▸ What Not to Edit Manually

Section	Reason
Authentication data (oauthAccount, primaryApiKey, customApiKeyResponses)	Risk of lock‑out or leaked secrets.
Application state (numStartups, cachedChangelog, …)	Non‑functional; overwritten by the app.
Projects block	Populated automatically and recalculated each session.

Expand the blocks only when debugging and restore from backup afterwards.

---

6 ▸ Validate & Reload

1. Validate JSON

   python -m json.tool ~/.claude.json
   # or
   jq . ~/.claude.json

2. Restart Claude Code

   claude

---

7 ▸ Common Tasks (Quick Checklist)

Task	Steps
Add new MCP server	Backup → Insert server block → Validate → Restart → /mcp to confirm
Change theme	Backup → Edit "theme" → Restart
Enable Vim mode	Backup → Set "editorMode": "vim" → Restart

---

8 ▸ Security Tips

• Keep ~/.claude.json private (chmod 600).
• Prefer environment variables for API keys over plain‑text.
• Never commit this file to source control.

---

Automation & Scripting Guide

Goal Show how to wire Claude Code into CI/CD pipelines and local Git hooks with verified, production‑tested snippets. All examples rely on Anthropic's public CLI (@anthropic-ai/claude-code

---

1 ▸ CI/CD Integration

1.1 GitHub Actions

name: Claude Code Review
on:
  pull_request:
    branches: [main, develop]

jobs:
  claude-review:
    runs-on: ubuntu-latest
    steps:
      - uses: actions/checkout@v4

      - name: Setup Node.js 18
        uses: actions/setup-node@v4
        with:
          node-version: '18'

      - name: Install Claude Code
        run: npm install -g @anthropic-ai/claude-code

      - name: Review PR
        env:
          ANTHROPIC_API_KEY: ${{ secrets.ANTHROPIC_API_KEY }}
        run: |
          claude -p "Review changes for security issues and bugs" \
            --allowedTools "View" \
            --output-format json > review-results.json

      - name: Upload results artifact
        uses: actions/upload-artifact@v4
        with:
          name: claude-review
          path: review-results.json

Key Points

Setting	Purpose
actions/checkout@v4	Retrieves the pull‑request diff.
@anthropic-ai/claude-code	Official CLI (auto‑updates disabled in CI for speed).
ANTHROPIC_API_KEY	Must be stored as an encrypted repo secret.
--allowedTools "View"	Read‑only toolset: prevents file writes in the runner.
--output-format json	Emits structured findings for downstream parsing.

Security tip: Restrict the runner's permissions (e.g. permissions: contents: read) so the CLI cannot push code back.

---

2 ▸ Local Git Automation

2.1 Pre‑commit Hook

#!/usr/bin/env bash
# .git/hooks/pre-commit (chmod +x)

# Abort if nothing staged
staged=$(git diff --cached --name-only --diff-filter=ACM)
[ -z "$staged" ] && exit 0

# Aggregate staged file contents
payload=$(echo "$staged" | xargs cat)

analysis=$(echo "$payload" | \
  claude -p "Review these changes for issues before commit" \
    --allowedTools "View" \
    --output-format json)

# Block commit on critical issues
if echo "$analysis" | jq -e '.critical_issues[]' >/dev/null 2>&1; then
  echo "❌ Critical issues found – commit blocked"
  exit 1
fi

echo "✅ Claude analysis passed"

Why This Works

• git diff --cached targets only staged changes, avoiding noise.
• xargs cat concatenates those files for the prompt.
• jq checks the JSON for a non‑empty critical_issues array.
• Hook exits non‑zero to stop the commit on failures.

⚠️ Performance note: For large diffs (>15 kB) invoke Claude with --stream to reduce latency.

---

3 ▸ Common Patterns

Use‑case	Flag combo	Example
Security review	--allowedTools "View"	claude -p "Audit for secrets" --allowedTools "View"
Auto‑fix (experimental)	--allowedTools "View,Write" --apply-patch	claude -p "Fix lint" --apply-patch
Generate SBOM	--allowedTools "View" --output-format cyclonedx	claude -p "Generate SBOM"

ℹ The --apply-patch flag is beta as of CLI v1.8. Check release notes before enabling in CI.

---

4 ▸ Best Practices

1. Rate limits — The free Anthropic tier caps at 100 requests/day. Cache results or run only on large PRs.
2. Timeouts — Use --timeout 120 to prevent hung CI jobs.
3. Artifact retention — Store review-results.json for traceability.
4. Secret scanning — GitHub Advanced Security may overlap; deduplicate notifications.

---

5 ▸ Troubleshooting

Symptom	Likely Cause	Fix
Error: Missing ANTHROPIC_API_KEY	Secret not set in repo or local env.	Define in Settings → Secrets or export locally.
CLI exits 1 with Rate limit exceeded	Too many calls in 24h.	Upgrade plan or throttle jobs.
Hook slow on binary files	Large payload sent to Claude.	Filter binary via git diff --cached --name-only --diff-filter=ACM -- '*.js' '*.ts'.

---

🔧 Troubleshooting

Diagnostic Commands

# Basic health checks
claude --version             
claude --help                   
claude config list                 
claude /doctor                  

Common Issues & Solutions

1. Authentication Issues

# Check API key
echo $ANTHROPIC_API_KEY

# Test connection
claude -p "test" --verbose

# Reset authentication 

2. Installation Issues

# Reinstall
npm uninstall -g @anthropic-ai/claude-code     
npm install -g @anthropic-ai/claude-code      

# Check Node.js version
node --version  # Should be 16+

3. Permission Issues

# Check current permissions
claude config get allowedTools

# Reset permissions
claude config set allowedTools "[]"
claude config set allowedTools '["Edit", "View"]'

4. MCP Issues

# Debug MCP 
claude --mcp-debug
claude mcp status  
claude mcp restart --all

Debug Mode

# Enable verbose logging
claude --verbose

# Check logs (verify log location)

---

🚀 Advanced Features

Context Management

CLAUDE.md - Project Memory

Purpose: Store persistent project information that Claude remembers

Location: Project root directory

# Project: My Application

## Overview
This is a React/Node.js application with PostgreSQL database.

## Architecture
- Frontend: React 18 with TypeScript
- Backend: Node.js with Express  
- Database: PostgreSQL 14

## Current Goals
- [ ] Implement authentication
- [ ] Add API documentation
- [ ] Set up CI/CD pipeline

## Development Guidelines
- Use TypeScript for all new code
- Follow ESLint configuration
- Write tests for new features

Memory Commands

claude /memory           # Edit project memory
claude /memory view      # View current memory

Advanced Thinking

# These "thinking" phrases may work:
claude "think hard about the security implications of this code"
claude "analyze this thoroughly and provide detailed recommendations"

Multi-Directory Workspaces

# Add multiple directories
claude --add-dir ../frontend ../backend ../shared

# Project-wide analysis
claude "analyze the entire application architecture"

---

💡 Best Practices

Effective Prompting

# Good: Specific and detailed
claude "Review UserAuth.js for security vulnerabilities, focusing on JWT handling"

# Bad: Vague  
claude "check my code"

Security Best Practices

1. Start with minimal permissions: claude --allowedTools "View"
2. Use environment variables: export API_KEY="secret"
3. Regular audits: claude config get allowedTools
4. Avoid dangerous mode: Only use --dangerously-skip-permissions in containers

Performance Tips

1. Use appropriate output formats: --output-format json for automation
2. Be specific in prompts: Better results, faster execution
3. Clean up regularly: Remove old sessions and cache

---

Last Updated: Based on package version 1.0.38 and available documentation. Many features require verification in your specific installation.

Monitoring & Alerting

1. Health Check Automation

# Regular health checks
*/15 * * * * /usr/local/bin/claude /doctor > /dev/null || echo "Claude health check failed" | mail -s "Alert" admin@company.com

2. Log Analysis

# Daily log analysis
0 6 * * * tail -1000 /var/log/app.log | claude -p "analyze for issues" --output-format json > /tmp/daily-analysis.json

Collaboration Best Practices

Team Workflows

1. Shared Configuration Templates

# Create team templates
mkdir -p ~/.claude/templates/
cat > ~/.claude/templates/team-frontend.json << EOF
{
  "allowedTools": ["Edit", "View", "Bash(npm:*)", "mcp__git__*"],
  "model": "claude-sonnet-4",
  "systemPrompt": "You are working on our React frontend. Follow our coding standards and use TypeScript."
}
EOF

# Use templates
claude config import ~/.claude/templates/team-frontend.json

2. Documentation Automation

# Automated documentation updates
claude "update README.md with recent changes to the API endpoints"
claude "generate TypeScript definitions from the new database schema"

3. Code Review Standards

# Standardized review process
claude --allowedTools "View,mcp__git__*" \
  "review PR #123 using our team standards:
  - Security best practices
  - Performance considerations  
  - Code style compliance
  - Test coverage adequacy"

Knowledge Sharing

1. Create Project Runbooks

# Generate runbooks
claude "create a deployment runbook for this application including all steps and troubleshooting"
claude "document the onboarding process for new developers"

2. Architecture Documentation

# Maintain architecture docs
claude "update architecture documentation to reflect recent microservices changes"
claude "create sequence diagrams for the new authentication flow"

Common Pitfalls to Avoid

Security Pitfalls

❌ Don't:

• Use --dangerously-skip-permissions on production systems
• Hardcode secrets in commands or configuration
• Grant overly broad permissions
• Run with elevated privileges unnecessarily

✅ Do:

• Use environment variables for secrets
• Start with minimal permissions
• Regular security audits
• Isolate sensitive operations

Performance Pitfalls

❌ Don't:

• Load entire large codebases unnecessarily
• Use maximum thinking budget for simple tasks
• Run multiple concurrent Claude instances
• Ignore memory and cache cleanup

✅ Do:

• Use focused context with --add-dir
• Match thinking budget to task complexity
• Monitor resource usage
• Clean up regularly

Workflow Pitfalls

❌ Don't:

• Skip project context setup (CLAUDE.md)
• Use vague, ambiguous prompts
• Ignore error messages and logs
• Automate without testing first

✅ Do:

• Maintain comprehensive project context
• Be specific and detailed in requests
• Monitor and analyze logs
• Test automation in safe environments

---