NativeWeather-Agent 🧠🌤️

A framework-free, first-principles implementation of an autonomous real-time weather agent loop utilizing Groq and LLaMA 3.3-70B.

🚀 Architectural Value (Why this exists)

Most modern AI applications abstract away core reasoning loops inside heavy third-party orchestration wrappers like LangChain or LangGraph. While convenient, this often obscures the underlying mechanics of context logging and model interaction.

NativeWeather-Agent completely eliminates this abstraction layer. Built entirely from scratch using native Python dictionaries, standard requests pipelines, and native Groq tool-calling schemas, this repository serves as a portfolio piece proving deep understanding of state log manipulation, autonomous runtime loops, routing logic, and dynamic tool dependencies.

🛠️ Multi-Step Dependency Execution

The agent natively resolves complex, multi-turn tool dependency cycles. For example, when processing a query like: "What is the weather in Bathinda right now?", the agent executes a sequential workflow completely decoupled from hardcoded logic:

1. Reasoning Step: Analyzes the query, recognizes a lack of real-time spatial knowledge, and schedules get_coordinates.
2. Context Logging: Appends the assistant's tool-call intent into the persistent message state array.
3. Dynamic Injection: Executes the native geocoding API request and pipes the resulting latitude/longitude outputs straight into a subsequent evaluation.
4. Tool Transition: Automatically triggers get_weather with the freshly parsed coordinates.
5. Synthesis: Grounds the finalized real-time meteorological vectors back into the LLM context window for a clean text summary.

🔒 Production Edge-Case Guardrails

• Zero Abstraction Bloat: Built using lightweight, native python dictionaries and native JSON schemas, maximizing inference tracking speed.
• Safe Attribute Reflection: Utilizes getattr(message, "tool_calls", None) validation to gracefully intercept terminal non-tool responses from the model, eliminating standard NoneType attribute crashes.
• Environment Isolation: Securely decouples API credentials from the code path using python-dotenv.

📦 Tech Stack

• Inference Model: LLaMA-3.3-70b-versatile (via Groq API)
• Core Language: Python 3.10+
• External APIs: Open-Meteo Geocoding & Weather Forecast Engines

🔧 Installation & Set Up

1. Clone the repository:

   git clone [https://github.com/your-username/NativeWeather-Agent.git](https://github.com/your-username/NativeWeather-Agent.git)
   cd NativeWeather-Agent
   

2. Install dependencies:

   pip install -r requirements.txt

3. Set up your API key: Create a .env file in the root directory and add your Groq API key:

   GROQ_API_KEY=your_groq_api_key_here
   

4. Run the agent:

   python agent.py