Skip to content

Installation

Multiple ways to install and run Remembra.

The easiest way to get started. Everything is bundled.

docker run -d \
  --name remembra \
  -p 8787:8787 \
  -e OPENAI_API_KEY=sk-your-key \
  -v remembra-data:/app/data \
  remembra/remembra

See Docker Guide for production configuration.

Python Package

SDK Only (Client)

If you just need the client SDK to connect to an existing Remembra server:

pip install remembra

Full Server

To run your own Remembra server:

pip install "remembra[server]"

Then start it:

export OPENAI_API_KEY=sk-your-key
python -m remembra.server

With Reranking (Optional)

For better recall quality with CrossEncoder reranking:

pip install "remembra[server,rerank]"

From Source

For development or customization:

# Clone the repo
git clone https://github.com/remembra/remembra
cd remembra

# Install with uv (recommended)
uv sync --all-extras

# Or with pip
pip install -e ".[server,rerank,dev]"

# Run tests
pytest

# Start the server
python -m remembra.server

Dependencies

Required

  • Python 3.10+
  • Qdrant - Vector database (bundled in Docker, or run separately)
  • OpenAI API key - For embeddings and extraction

Optional

  • Ollama - Local embeddings (no API costs)
  • Cohere - Alternative embeddings
  • Redis - For rate limiting at scale

Embedding Providers

Remembra supports multiple embedding providers:

export OPENAI_API_KEY=sk-your-key
export REMEMBRA_EMBEDDING_PROVIDER=openai
export REMEMBRA_EMBEDDING_MODEL=text-embedding-3-small

# Start Ollama first
ollama pull nomic-embed-text

export REMEMBRA_EMBEDDING_PROVIDER=ollama
export REMEMBRA_EMBEDDING_MODEL=nomic-embed-text
export OLLAMA_BASE_URL=http://localhost:11434

export COHERE_API_KEY=your-key
export REMEMBRA_EMBEDDING_PROVIDER=cohere
export REMEMBRA_EMBEDDING_MODEL=embed-english-v3.0

Verifying Installation

Check Server Health

curl http://localhost:8787/health

Expected response:

{
  "status": "healthy",
  "version": "0.6.3",
  "qdrant": "connected",
  "database": "connected"
}

Test the SDK

from remembra import Memory

memory = Memory(
    base_url="http://localhost:8787",
    user_id="test"
)

# Store and recall
memory.store("Test memory")
result = memory.recall("test")
print(result)  # Should return the test memory

Troubleshooting

"Connection refused" error

Make sure the server is running:

docker ps  # Check if container is running
docker logs remembra  # Check for errors

"API key not set" error

Set your OpenAI API key:

export OPENAI_API_KEY=sk-your-key

Qdrant connection issues

If running Qdrant separately, ensure it's accessible:

export QDRANT_HOST=localhost
export QDRANT_PORT=6333

Next Steps