AI Node Analyzer

AI-powered blockchain node log analyzer that detects sync failures, peer connectivity issues, and infrastructure bottlenecks.

---

Features

• Detects peer connection failures
• Identifies blockchain sync stalls
• Flags disk space warnings
• Provides technical remediation suggestions
• Returns structured JSON output with severity levels

---

Installation

Install dependencies:

pip install -r requirements.txt

Create a .env file in the project root:

OPENAI_API_KEY=your_api_key_here

Windows Quick Start

Tested on Windows with Git and Python installed.

1. Clone the repository:

   git clone https://github.com/CMZS4/ai-node-analyzer.git
   cd ai-node-analyzer

2. (Optional) Create a virtual environment:

   python -m venv .venv
   .venv\Scripts\activate

3. Install dependencies:

   pip install -r requirements.txt

4. Create .env and add your OpenAI API key:

   OPENAI_API_KEY=your_api_key_here

5. Run:

   python app.py

---

Usage

Place your node log file as:

node_log.txt (or use the sample in examples/node_log.sample.txt)

Run the analyzer:

python app.py

Results are saved to:

analysis_result.json

---

Tech Stack

• Python
• OpenAI API
• python-dotenv

---

Roadmap

• CLI arguments (--input, --output)
• Improved error handling
• Web dashboard (Streamlit)
• Docker support

Example Files

Sample files are included in the examples/ directory:

• examples/node_log.sample.txt
• examples/analysis_result.sample.json

---

CLI Usage Examples

Run with default paths:

python app.py

Run with a custom input file: 
python app.py --input examples/node_log.sample.txt

Run with a custom output file:

python app.py --input examples/node_log.sample.txt --output examples/custom_result.json

Troubleshooting
OPENAI_API_KEY is missing

Create a .env file in the project root and add:

OPENAI_API_KEY=your_api_key_here

Log file not found

Make sure the input path exists, or use:

python app.py --input path/to/your/node_log.txt

JSON parsing failed

Sometimes model output may not be valid JSON. In that case, the raw response is saved to:

raw_model_output.txt

You can inspect it and retry with a shorter or cleaner log input.