Skip to content

README.md

Latest commit

 

History

History
203 lines (138 loc) · 5.44 KB

README.md

File metadata and controls

203 lines (138 loc) · 5.44 KB

Claude Code Agent Reference Documentation

Version: 1.0.0 Last Updated: 2025-12-26 Based on: Anthropic Official Documentation

Overview

This documentation provides a comprehensive reference guide for developing AI Agents using Claude Code. Based on Anthropic's official documentation, it covers core concepts including Agent SDK, tools, hooks, MCP integration, and more.

Table of Contents

Fundamentals

# Document Description
00 Overview Claude Code Agent introduction and architecture
01 Agent SDK Agent SDK installation, setup, and core APIs
02 Tools Reference Built-in tools (Read, Edit, Bash, etc.) details

Extensions

# Document Description
03 Hooks Guide Customizing agent behavior with hooks
04 MCP Integration External system integration (databases, APIs)
05 Skills Reference Defining reusable modular capabilities

Operations

# Document Description
06 Configuration Guide Permissions, environment variables, model settings
07 Security Guide Isolation, credential protection, auditing
-- Headless Execution CI/CD integration and automation scripts

Advanced

# Document Description
08 Architecture Patterns Single/multi-agent, workflow patterns
09 API Reference Messages API, Agent SDK API details

Quick Start

1. Installation

# Install Claude Code CLI
npm install -g @anthropic-ai/claude-code

# Install Agent SDK (Python)
pip install claude-agent-sdk

# Install Agent SDK (TypeScript)
npm install @anthropic-ai/claude-agent-sdk

2. Authentication

export ANTHROPIC_API_KEY=your-api-key

3. Your First Agent

import asyncio
from claude_agent_sdk import query, ClaudeAgentOptions

async def main():
    async for message in query(
        prompt="Analyze the code in the src folder",
        options=ClaudeAgentOptions(
            allowed_tools=["Read", "Glob", "Grep"]
        )
    ):
        print(message)

asyncio.run(main())

Learning Path

Beginners

  1. Overview - Understand the overall structure
  2. Agent SDK - Basic usage
  3. Tools Reference - Working with built-in tools

Intermediate

  1. Hooks Guide - Customizing behavior
  2. Configuration Guide - Permissions and settings
  3. MCP Integration - External system integration

Advanced

  1. Security Guide - Production security
  2. Architecture Patterns - Complex system design
  3. API Reference - Detailed API information

Use Case References

Building a Code Analysis Agent

Building a Code Modification Agent

External System Integration

Production Deployment

Key Concepts Summary

Agent Execution Flow

Prompt → Claude selects tools → SDK executes tools → Analyze results → Repeat/Complete

Main Components

Component Role
Agent SDK Agent loop and tool execution
Tools File, command, web operations
Hooks Lifecycle event interception
MCP External system integration
Skills Reusable capability modules
Permissions Tool access control

Permission Modes

Mode Description
default Confirm all actions
acceptEdits Auto-approve file edits
bypassPermissions Bypass all permissions (dangerous!)
plan Plan only, no execution

Official Resources

Resource URL
Agent SDK Docs https://platform.claude.com/docs/en/agent-sdk/overview
Claude Code Docs https://code.claude.com/docs/en/overview
Tool Use Guide https://platform.claude.com/docs/en/agents-and-tools/tool-use/overview
MCP Protocol https://modelcontextprotocol.io
API Reference https://docs.anthropic.com/en/api
Community MCP Servers https://github.com/modelcontextprotocol/servers

Language Versions

Version History

Version Date Changes
1.0.0 2025-12-26 Initial documentation

Contributing & Feedback

Please file an issue for improvement suggestions or bug reports.

To get started with Claude Code Agent development, begin with the Overview.