Prompt-Engineering--Poject

🤖 Natural Language to CLI Agent (MVP) An intelligent translator that converts natural language instructions into precise terminal commands. This project focuses on Prompt Engineering and Safety Guardrails to ensure that AI-generated commands are accurate and secure before execution.

🌟 Key Features Multi-Platform Support: Generates syntax for Windows (CMD), Linux, and macOS.

Safety Classification System: Implements a 3-tier safety logic:

✅ SAFE: Informational or non-destructive commands.

⚠️ DANGEROUS: Potential data-loss operations (tagged for user approval).

🚫 FORBIDDEN: Blocks catastrophic or system-damaging commands (e.g., format).

Context Awareness: Automatically leverages system environment variables (like %USERPROFILE%) and ensures proper path quoting for high reliability.

User-Friendly Interface: Built with Gradio for seamless interaction.

🧠 Prompt Engineering Strategy The core of this project is a sophisticated System Prompt that enforces strict output rules:

Zero-Shot Classification: Categorizing commands in real-time without prior training data.

Strict Format Control: Ensuring the LLM returns only the command string, preventing conversational "noise" that would break a terminal execution.

Error Handling: Built-in mechanisms to catch API exceptions and provide clean user feedback.

🛠️ Tech Stack Language: Python

AI Engine: OpenAI GPT-4o-mini

Frontend/UI: Gradio

Environment: Dotenv for secure API key management

⚙️ Installation & Usage Clone the repository:

Install dependencies:

Configure Environment: Create a .env file in the root directory and add your key:

Run the Application:

📊 Evaluation & Metrics The system was tested against various Edge Cases (ambiguous instructions, dangerous commands, system-specific syntax) to achieve high accuracy and safety reliability. Through iterative prompt optimization, the agent ensures that dangerous operations are never presented as "safe".