Claude Code Feature Reference

Comprehensive list of Claude Code features as of January 2026.

Configuration Files

Hooks (13 Types)

Hook Types

• command - Execute shell command
• prompt - Inject prompt (Stop/SubagentStop only)

Hook Response Format

{
  "continue": true,
  "stopReason": "string",
  "systemMessage": "string",
  "hookSpecificOutput": {
    "hookEventName": "PreToolUse",
    "permissionDecision": "allow|deny|ask"
  }
}

Skills

Required Frontmatter

• name - Lowercase, hyphens, max 64 chars
• description - When to use, max 1024 chars

Optional Frontmatter

• allowed-tools - Tools allowed without permission
• model - Specific model to use
• context: fork - Isolated sub-agent context
• agent - Agent type for fork context
• hooks - Lifecycle hooks
• user-invocable - Slash command visibility

Subagents

Required Frontmatter

• name - Lowercase with hyphens
• description - When to delegate

Optional Frontmatter

• tools - Available tools
• disallowedTools - Denied tools
• model - sonnet, opus, haiku, inherit
• permissionMode - default, acceptEdits, dontAsk, bypassPermissions, plan
• skills - Skills to load
• hooks - Lifecycle hooks

Built-in Agents

• Explore - Fast read-only (Haiku)
• Plan - Research for planning
• general-purpose - Complex tasks
• Bash - Terminal commands

Plugins

Structure

plugin/
├── .claude-plugin/plugin.json
├── commands/
├── agents/
├── skills/
├── hooks/hooks.json
├── .mcp.json
└── .lsp.json

Permissions

{
  "permissions": {
    "allow": ["Bash(git diff:*)"],
    "ask": ["Bash(git push:*)"],
    "deny": ["Read(./.env)"]
  }
}

MCP Servers

{
  "mcpServers": {
    "name": {
      "command": "npx",
      "args": ["-y", "@package/name"],
      "env": {}
    }
  }
}

Last verified: January 2026