Installation¶

Muffakir 0.3.0 supports Python 3.11 and later. Its default package is intentionally small; install only the capabilities your application uses.

Choose an installation profile¶

# Public API, configuration, and prompt management
pip install Muffakir

# Local RAG, Composer, and ComposerUI
pip install "Muffakir[standard]"

# ComposerUI server without local ML/vector/database integrations
pip install "Muffakir[ui]"

# Every supported integration
pip install "Muffakir[all]"

For a smaller production environment, combine feature extras:

pip install "Muffakir[rag,local,chroma,openai]"
pip install "Muffakir[datasets,pdf,token,bm25]"

Need	Extra
Local sentence-transformer embeddings and Hugging Face rerankers	`local`
Chroma vector database	`chroma`
OpenAI models	`openai`
ComposerUI server	`ui`
Tabular evaluation datasets	`datasets`
BM25 reranking	`bm25`

Development installation¶

git clone https://github.com/Mohamed28112003/Muffakir.git
cd Muffakir
python -m venv .venv

# Windows
.venv\Scripts\activate

# Linux/macOS
source .venv/bin/activate

pip install -r requirements.txt

Optional dependencies¶

Muffakir validates optional features when you configure them. If a capability is missing, it raises MissingOptionalDependencyError and tells you which extra to install. It never installs packages automatically at runtime.

Next: build your first RAG application.