AutoRAG-Research 0.0.1

Automate your RAG research.

AutoRAG-Research

Automate your RAG research.

• Github repository: https://github.com/vkehfdl1/AutoRAG-Research/
• Documentation https://vkehfdl1.github.io/AutoRAG-Research/

Recommended Setup

The easiest way to get started is with the installation script:

curl -LsSf https://raw.githubusercontent.com/NomaDamas/AutoRAG-Research/main/scripts/install.sh -o install.sh
bash install.sh

The script will guide you through:

1. Setting up a Python environment (supports uv, venv, conda, pyenv, etc.)
2. Installing AutoRAG-Research
3. Configuring PostgreSQL (Docker or existing server)

CLI Usage

AutoRAG-Research provides a CLI tool for managing RAG research workflows.

Installation

pip install autorag-research

or

uv pip install autorag-research

Quick Start

# 1. Initialize configuration files
autorag-research init

# 2. Edit database settings
vim configs/db.yaml # OR your preferred editor

# 3. Ingest a dataset
autorag-research ingest --name beir --extra dataset-name=scifact

# 4. Run experiments
autorag-research run --db-name=beir_scifact_test

Commands

init - Initialize Configuration Files

Downloads default configuration files to ./configs/ directory.

autorag-research init

This creates:

• configs/db.yaml - Database connection settings
• configs/experiment.yaml - Experiment configuration
• configs/pipelines/**/*.yaml - Pipeline configurations
• configs/metrics/**/*.yaml - Metric configurations

ingest - Ingest Datasets

Ingest datasets into PostgreSQL. Each ingestor supports different datasets.

# Show available ingestors
autorag-research ingest --help

autorag-research ingest --name beir --embedding-model mock --query-limit 5 --min-corpus-cnt 10 --extra dataset-name=scifact

list - List Available Resources

# List available ingestors
autorag-research list ingestors

# List available pipelines
autorag-research list pipelines

# List available metrics
autorag-research list metrics

# List database schemas
autorag-research list databases

run - Run Experiments

Run experiment pipelines with metrics evaluation. Requires --db-name to specify the target database schema.

# Basic run (uses configs/experiment.yaml)
autorag-research run --db-name=beir_scifact_test --verbose

Environment Variables

Variable	Description
POSTGRES_PASSWORD	PostgreSQL password (recommended for security)
AUTORAG_CONFIG_PATH	Default configuration directory path

Implementing New Pipelines (with Claude Code)

This project includes specialized Claude Code agents for implementing new RAG pipelines from research papers.

Quick Start

# Full workflow from paper to validated code
/implement-pipeline https://arxiv.org/abs/2212.10496