Md Agent Pro — AI skill for Claude Code

Markdown Agent Pro

Learn LLM agent fundamentals by building from scratch

A minimal, standalone CLI agent demonstrating the core concepts of autonomous AI systems. Built with OpenAI's tool calling API, this project shows how simple file operations can enable powerful automation.

**Status**: Educational prototype - perfect for understanding agent fundamentals **Core Insight**: A CLI agent that can read, write, and edit files is one step away from total automation

Built with DeepSeek in Claude Code over a weekend, this project draws inspiration from Gemini-Cli, Claude Code, and Cursor.

🚀 Quick Start

Installation & Setup

git clone 
cd md_agent_pro
python3 -m venv .venv
source .venv/bin/activate
pip install -r requirements.txt
cp .env.example .env

Environment Configuration

API_KEY=your_openai_api_key_here
BASE_URL=https://api.openai.com/v1
MODEL=gpt-4-turbo
# Optional: Custom model configuration
# TEMPERATURE=0.3
# Edit .env with your OpenAI API key or any other Compatible LLM provider, I use DeepSeek.

Test Your Setup

Start with a simple command:

python -m src.cli "Read the first few lines and summarize" --file README.md

Try creative tasks:

python -m src.cli "Write a poem about the grace of machines"

Or demonstrate complex workflows without predefined steps:

python -m src.cli "Write a story about Bob in A.md and Alice in B.md, combine them in C.md, translate to Chinese in D.md, and summarize in E.md"

🏗️ Architecture & Core Features

Essential Components

• Generic Agent: Tool-agnostic framework with dynamic loading
• Filesystem Tools: Read, write, edit, grep, and glob operations
• Security: Workspace restrictions and path validation
• Reliability: Error recovery and completion detection

Project Structure

src/
├── core/           # Agent framework and tool management
├── tools/          # Filesystem operation tools
└── utils/          # Log