LLM-LOGR

LLM-LOGR is a lightweight utility for tracing, logging, and viewing LLM (Large Language Model) API calls locally.
It provides structured logging of inputs, outputs, latency, and metadata for OpenAI API interactions.

Features

• Log OpenAI API requests and responses
• JSON-based logging format
• Streamlit interface for viewing logged calls
• Local-first design with no external dependencies

Setup

1. Clone the repository:

git clone 
cd llm-logr

2. Create and activate a virtual environment:

python3 -m venv logr-venv
source logr-venv/bin/activate  # Windows: logr-venv\Scripts\activate

3. Install the requirements:

pip install -r requirements.txt

4. Install the package locally:

pip install -e .

5. Add your OpenAI API key in a .env file you created in the root directory:

OPENAI_API_KEY=your-openai-key-here

Usage

Log an OpenAI call

from llm_logr.core.log_openai_call import track
import openai

response = track(
    openai.ChatCompletion.create,
    user_id="u123",
    model="gpt-3.5-turbo",
    messages=[{"role": "user", "content": "Explain ISO 42001 in one line"}]
)

Logged results are saved in logs/llm_logs.json.

View logged calls

Run the Streamlit app:

streamlit run llm_logr/web/app.py

Development Notes

• Update requirements.txt by running:

pip freeze > requirements.txt

• Make sure .env, logr-venv/, and llm_logr.egg-info/ are listed in .gitignore.