RAG Debugger SDK

Observability & Debugging Platform for RAG Pipelines

ragdebug allows you to auto-trace every stage of your RAG application and start a comprehensive local dashboard for analysis with a single command.

Prerequisites

To run the full local RAG Debugger platform, you must have Docker Desktop installed and running on your machine.

• Download and Install Docker Desktop here

Once installed and running, you can follow the instructions below.

Quickstart

1. Install the SDK with the CLI power-ups:

   pip install "ragdebug[cli]"
   

2. Start the local RAG Debugger platform:

   ragdebug up
   

3. Open the dashboard at http://localhost:8000

Integrating Tracing

Add the @trace decorator to your RAG pipeline functions:

from ragdebug import init, trace, Prompt

init(project="my-project")

test_prompt = Prompt(name="QA Prompt", template="Context: {{context}}\n\nQ: {{query}}")

@trace(name="qa_pipeline")
def answer_question(query: str):
    # Your RAG logic here
    return "Answer"

answer_question("What is RAG?")

Advanced CLI Usage

Documentation

For full documentation on Prompts, Evaluations, and more, check the UI dashboard after running ragdebug up.