🔍 LLM Debugger
A Local-first Tool For Logging and Visualizing AI Conversations
🌐 View on GitHub
pip install llm-logger
What is it?
LLM Debugger is a local-first tool for logging and visualizing OpenAI API calls (Anthropic coming soon!). It helps you debug conversations and agents by logging everything locally, automatically grouping related messages, and rendering them in a simple UI. Check out the repository for more details.
Features
- 🧠 Prompt & response diffs between turns
- 📂 Logs stored locally as JSON
- ⚡ One-line setup via client wrapper
- 🔐 Fully offline — no server, no account
- 🔍 Automatic conversation tagging
- ⏱️ Metadata + latency tracking
- 🖥️ Simple, static UI served locally
- 🧩 Works with any Python codebase
Why use this?
Most LLM dev tools are hosted or heavyweight. LLM Debugger gives you a fast, lightweight way to understand what your model is doing — no cloud needed.
Get Started
Install the package (available on PyPI):
pip install llm-loggerThen launch the viewer:
llm_loggerOr embed the UI in your app using FastAPI or Flask.
🧪 Want to shape v2? Join the discussion on GitHub
🌐 View the Project on GitHub