Update README.md

README.md

@@ -1,13 +1,86 @@
 ---
-title: C
-emoji: 👀
+title: C-Codestral Agent
+emoji: 🤖
 colorFrom: green
 colorTo: gray
 sdk: gradio
-sdk_version: 5.33.0
+sdk_version: 4.31.0 # langchain과 호환성을 위해 4.x 버전을 명시하는 것이 안정적일 수 있습니다.
 app_file: app.py
-pinned: false
+pinned: true
 license: mit
+tags:
+  - agent-demo-track
+  - mcp-server-track
 ---
 
-Check out the configuration reference at https://huggingface.co/docs/hub/spaces-config-reference
+# 🚀 C-Codestral Agent: Your C-Language Co-pilot
+
+**C-Codestral Agent** is a powerful, conversational AI agent designed to assist developers with C programming. Powered by Mistral's `Codestral` model and built with Gradio and LangChain, this agent can generate, analyze, refactor, compile, and run C code seamlessly within a single interface.
+
+This project is a submission for the **Agents & MCP Hackathon 2025**, demonstrating the power of AI agents and the Model Context Protocol (MCP).
+
+[![Gradio App](https://img.shields.io/badge/Gradio-App-blue)](https://huggingface.co/spaces/Agents-MCP-Hackathon/c) <!-- 이 부분은 실제 Space URL로 자동 연결됩니다 -->
+
+---
+
+## 🎥 Demo Video
+
+Here is a short video demonstrating the core functionalities of the C-Codestral Agent, including code generation, compilation, and its use as an MCP tool.
+
+**(ここにデモビデオへのリンクを挿入してください。例: [Watch the Demo on YouTube](https://youtu.be/your_video_id))**
+
+*중요: 데모 영상 링크를 여기에 꼭 추가하세요. Loom, YouTube, 또는 Space에 직접 업로드한 영상 링크도 좋습니다.*
+
+---
+
+## ✨ Core Features
+
+*   **🤖 Conversational Code Generation**: Simply ask the agent in natural language to create C code for any task, from simple "Hello, World!" programs to complex algorithms.
+*   **✅ Instant Compilation & Execution**: Test the generated or provided C code immediately. The agent uses a secure backend with `gcc` to compile and run the code, showing you the output or any errors in real-time.
+*   **🧠 Intelligent Code Analysis & Refactoring**: Provide your existing C code and ask for improvements, optimizations, or explanations. The agent helps you write better, more efficient code.
+*   **🛠️ MCP-Enabled Tools**: The agent's core capabilities (generate, compile, refactor) are exposed as MCP-compliant tools. This allows other LLM agents (like Claude Desktop, Cursor, etc.) to use this Gradio Space as a powerful C programming "tool" to solve complex problems.
+
+---
+
+## 🛠️ Tech Stack & Architecture
+
+This project integrates several cutting-edge technologies to create a robust agentic system:
+
+*   **LLM (The Brain)**: `codestral-latest` by Mistral AI for its state-of-the-art code generation and understanding capabilities.
+*   **Agent Framework**: `LangChain` to orchestrate the agent's logic, manage conversation history, and intelligently decide which tool to use.
+*   **Frontend & UI**: `Gradio` for creating a sleek, responsive, and interactive chat interface.
+*   **Backend & Execution Environment**: Hugging Face Spaces, utilizing a secure Linux container with `gcc` for C code compilation.
+*   **Tool Protocol**: Model Context Protocol (MCP) via Gradio's `@tool` and `gr.load_tools()` functionalities, making the agent's skills available to the wider AI ecosystem.
+
+<p align="center">
+  <img src="https://i.imgur.com/your_architecture_diagram_url.png" width="700">
+  <em>(선택 사항: 아키텍처 다이어그램 이미지를 추가하면 전문성을 더할 수 있습니다.)</em>
+</p>
+
+---
+
+## 🚀 How to Use
+
+1.  **Chat with the Agent**: Open the Gradio app and start a conversation. You can ask it to:
+    *   `"Generate a C program to find all prime numbers up to 100."`
+    *   `(After code is generated) "Now, compile and run it."`
+    *   `(Paste your code) "Refactor this code to be more memory-efficient."`
+
+2.  **Use as an MCP Tool**:
+    *   Connect your favorite MCP client (e.g., a compatible version of Claude Desktop).
+    *   Add this Hugging Face Space as a tool.
+    *   You can now ask the client to perform C-related tasks, and it will use this agent's tools (`generate_c_code`, `compile_and_run_c_code`, etc.) to get the job done.
+
+---
+
+## 🏆 Hackathon Submission Details
+
+This project is submitted for the following tracks and awards:
+
+*   **Primary Track**: `agent-demo-track`
+*   **Secondary Track**: `mcp-server-track`
+*   **Hoping to be considered for**:
+    *   Most Innovative Use of MCP Award
+    *   Community Choice Award
+
+Thank you for checking out our project!