Installation

Ragrails requires Python 3.10 or newer and installs with pip.

pip install ragrails

pip install ragrails
ragrails --help

pip install "ragrails[server-qdrant]"
ragrails-api --help

The base install includes the SDK, CLI, REST API command, OpenAI, Anthropic, Google, document ingestion, API ingestion, chunking, embedding orchestration, vector storage orchestration, retrieval, chat, and pipeline helpers.

OpenAI, Anthropic, and Google clients are included in the base package. Set the matching API key before using the provider.

Common installs

Choose the smallest install that matches the thing you are running.

Goal	Install
SDK, CLI, docs/API ingestion, chunking, retrieval, chat	`pip install ragrails`
SDK pipeline with Voyage + local or cloud Qdrant	`pip install "ragrails[store-qdrant]"`
SDK pipeline with Voyage + Pinecone	`pip install "ragrails[store-pinecone]"`
SDK pipeline with Voyage + Weaviate	`pip install "ragrails[store-weaviate]"`
REST API backed by Qdrant	`pip install "ragrails[server-qdrant]"`
REST API backed by Pinecone	`pip install "ragrails[server-pinecone]"`
REST API backed by Weaviate	`pip install "ragrails[server-weaviate]"`
Website scraping	`pip install "ragrails[url]"`
Local reranking	`pip install "ragrails[rerank]"`
Everything	`pip install "ragrails[all]"`

For website scraping, install the browser runtime once in the environment that will scrape pages:

ragrails setup-url --browser chromium

from ragrails import RagRails

rag = RagRails()
rag.setup_url(browser="chromium")

Extras

Extras add integrations that are optional at runtime. Combine extras with commas when you need more than one.

Extra	Adds
`url`	Crawl4AI website scraping support
`voyage`	Voyage embeddings and Voyage reranking credentials/client support
`qdrant`	Qdrant vector database client
`pinecone`	Pinecone vector database client
`weaviate`	Weaviate vector database client
`rerank`	Local reranking dependencies
`store-qdrant`	Voyage + Qdrant for SDK/CLI storage pipelines
`store-pinecone`	Voyage + Pinecone for SDK/CLI storage pipelines
`store-weaviate`	Voyage + Weaviate for SDK/CLI storage pipelines
`server`	REST server dependencies only
`server-qdrant`	REST server + Voyage + Qdrant
`server-pinecone`	REST server + Voyage + Pinecone
`server-weaviate`	REST server + Voyage + Weaviate
`all`	All optional integrations

Combined extras

pip install "ragrails[url,store-qdrant,rerank]"

Provider keys

Set credentials in environment variables, not in docs examples, source files, or .ragrails.toml.

OpenAI chat + Voyage embeddings

export OPENAI_API_KEY="sk-..."
export VOYAGE_API_KEY="pa-..."

Vector database keys

export QDRANT_API_KEY="..."
export PINECONE_API_KEY="..."
export WEAVIATE_API_KEY="..."

Qdrant Cloud uses the same qdrant, store-qdrant, and server-qdrant extras as local Qdrant. Point the provider URL at your cloud endpoint and set QDRANT_API_KEY.

Local Qdrant

The quickstart path uses local Qdrant:

Docker

docker run -p 6333:6333 qdrant/qdrant

Verify

Check the import, CLI, and REST command before building a pipeline:

python -c "from ragrails import RagRails; print(RagRails.__name__)"

ragrails --help

ragrails-api --help

Then continue with the quickstart, environment variables, or vector databases.

Get Started

Features

Capabilities

Use Cases

Walkthroughs

Common installs

Extras

Provider keys

Local Qdrant

Verify

​Common installs

​Extras

​Provider keys

​Local Qdrant

​Verify

Common installs

Extras

Provider keys

Local Qdrant

Verify