Installation

Requirements

• Python 3.11+ (3.11–3.13 wheels published; see Python SDK)
• macOS or Linux (Windows not yet supported)
• No C compiler required on supported platforms — the engine ships inside the wheel

Install from PyPI

pip install hypermesh

Optional extras:

pip install "hypermesh[analytics]"     # hypergraph analytics (numpy/scipy)
pip install "hypermesh[server]"        # REST API server (hmdb serve)
pip install "hypermesh[virtual]"       # HyperMesh Virtual (warehouse queries)
pip install "hypermesh[mcp]"           # MCP verifiable-memory server
pip install "hypermesh[all]"           # everything

Backward compatibility

hypermesh is the PyPI distribution name. After install, import hypermeshdb continues to work as a legacy import alias in the same wheel — see the migration guide.

Prebuilt wheels: CPython 3.11–3.13 on Linux manylinux2014 (x86_64 + aarch64) and macOS arm64. Other platforms install from sdist (requires gcc/make) or use the remote client only.

Install from source

git clone https://github.com/sree181/hypermesh-workbench.git
cd hypermesh-workbench
pip install -e ".[dev]"

This installs in editable mode with the full development toolchain (pytest, ruff, mypy, FastAPI server deps).

Verify the installation

hmdb --version
python3 -c "import hypermesh as hm; print(hm.__version__)"

import hypermesh as hm

db = hm.connect("/tmp/verify_install")
print(db.execute("CALL SHOW_HYPEREDGE_TABLES() RETURN *"))

TypeScript / Node.js SDK

npm install @hypermeshdb/client

Requires Node.js 18 or later (for native fetch).

Docker

docker pull hypermeshdb/server:latest
docker run -p 8000:8000 \
  -v /your/data:/data \
  -e HMDB_API_KEY=hmdb_yoursecretkey \
  hypermeshdb/server

See the Docker guide for full compose configuration with TLS, volumes, and Kubernetes readiness/liveness probes.