An educational project that teaches you how to build modern AI coding agents from scratch through progressive tutorials, featuring 5 versions from simple bash tools to a complete skills system.
One-Minute Overview#
This is an educational project that teaches you how to build AI coding agents similar to Claude Code from scratch through 5 progressive versions (from about 50 to 550 lines of code). Whether you want to understand how AI agents work or build your own agent system, this project provides a clear path with practical code examples.
Core Value: Through progressive code examples, reveals the essential design patterns of AI agents, helping you understand the core concept of "model as agent".
Quick Start#
Installation Difficulty: Low - Simple pip installation and configuration
# Install dependencies
pip install -r requirements.txt
# Configure API
cp .env.example .env
# Edit .env with your API key (supports Anthropic, OpenAI, Gemini, etc.)
# Run any version
python v0_bash_agent.py # Minimal version
python v1_basic_agent.py # Core agent loop
python v2_todo_agent.py # + Todo planning
python v3_subagent.py # + Subagents
python v4_skills_agent.py # + Skills system
Is this suitable for my scenario?
- ✅ AI Agent Development Learning: Want to understand how AI agents work and their design patterns
- ✅ Educational Research: Need progressive examples to teach AI agent concepts
- ❌ Production Deployment: This is an educational project, not a production-ready solution
- ❌ Rapid Integration: Doesn't provide plug-and-play components, but rather a complete learning experience
Core Capabilities#
1. Progressive Learning Path - Simple to Complex#
- Builds complexity through 5 versions (v0-v4), each adding a key concept Actual Value: Allows learners to understand the evolution of AI agents stage by stage, avoiding being overwhelmed by complex concepts all at once
2. Core Agent Pattern Revelation - Revealing the Essence of AI Agents#
- Shows the basic loop of all AI agents: model calls tools → execute tools → repeat until done Actual Value: After understanding this simple loop, you can quickly analyze and understand how various AI agents work
3. Skills System Implementation - Providing Domain Expertise#
- Shows how to provide on-demand domain expertise to agents through SKILL.md files Actual Value: Enables agents to acquire domain knowledge dynamically based on task requirements, not just relying on pre-trained knowledge
4. Agent Builder Tool - Accelerating Project Initialization#
- Provides a meta-skill to quickly set up new agent project frameworks Actual Value: Gives developers a standardized agent project initialization process, reducing repetitive work
Technical Stack & Integration#
Development Language: Python Main Dependencies: Managed through requirements.txt, supports multiple APIs (Anthropic, OpenAI, Gemini, etc.) Integration Method: Codebase/tutorial, can run as standalone project or serve as reference template
Maintenance Status#
- Development Activity: Actively maintained with clear version progression path
- Recent Updates: Recently updated with complete documentation and examples
- Community Response: Has related skills library and specification projects, indicating community support
Commercial & License#
License: MIT
- ✅ Commercial: Allowed
- ✅ Modification: Allowed
- ⚠️ Restrictions: Must include original copyright and license notices
Documentation & Learning Resources#
- Documentation Quality: Comprehensive
- Official Documentation: docs/ directory contains English and Chinese tutorials
- Example Code: 5 complete working versions (v0-v4), ranging from 50 to 550 lines