Community preview / 1.0.1rc4

Deploy without repository access.

The AGPL Community release candidate is prepared for PyPI and versioned direct distribution from lummenna.com.

01 / Artifacts

Versioned and verifiable.

Verify every direct artifact against the versioned checksum manifest before installation.

Runtime package

Python wheel

Python 3.10 through 3.12. Its digest is recorded in the release manifest.

Download wheel

Corresponding source

Source archive

Python and React source, build metadata, benchmark artifacts, notices, and licenses.

Download source

Download sidecar Dockerfile / Download SHA256SUMS-1.0.1rc4

Choose a toolchain

Four locked paths, one Community release.

uv installs the Python runtime and MCP. npm, pnpm, and Bun build the React frontend from the corresponding source archive.

02. Download Community with curl

This URL contains only the AGPL Community wheel.

curl
$ curl --fail --location --show-error https://lummenna.com/downloads/lummenna-1.0.1rc4-py3-none-any.whl --output lummenna-1.0.1rc4-py3-none-any.whl
$ curl --fail --location --show-error https://lummenna.com/downloads/SHA256SUMS-1.0.1rc4 --output SHA256SUMS-1.0.1rc4
$ sha256sum --check --ignore-missing SHA256SUMS-1.0.1rc4
ready

03. Install the preview with MCP

uv or pip
$ uv tool install --python 3.12 "lummenna[mcp,export]==1.0.1rc4"
$ python -m pip install "./lummenna-1.0.1rc4-py3-none-any.whl[mcp,export]"
ready

04. Build the sidecar

docker build
$ curl --fail --show-error --silent --remote-name https://lummenna.com/downloads/lummenna-community-1.0.1rc4.Dockerfile
$ docker build --pull --file lummenna-community-1.0.1rc4.Dockerfile --tag lummenna-community:1.0.1rc4 .
ready

05. Provision the model once

model volume
$ docker volume create lummenna-models
$ docker run --rm --init --read-only --cap-drop ALL --security-opt no-new-privileges=true --pids-limit 256 --tmpfs /tmp:rw,nosuid,size=2g --env-file lummenna.env --env LUMMENNA_SOVEREIGN=false --volume lummenna-models:/data/models lummenna-community:1.0.1rc4 lummenna model download bge-small-en-v1.5
ready

06. Run beside an agent

hardened sidecar
$ docker network create agent-memory
$ docker volume create lummenna-store
$ docker run --detach --name lummenna --init --read-only --restart unless-stopped --cap-drop ALL --security-opt no-new-privileges=true --pids-limit 256 --tmpfs /tmp:rw,noexec,nosuid,size=64m --network agent-memory --env-file lummenna.env --publish 127.0.0.1:8848:8848 --volume lummenna-store:/data/store --volume lummenna-models:/data/models lummenna-community:1.0.1rc4
ready

07. Run the MCP sidecar over stdio

The same image includes the MCP extra. Keep networking disabled when the MCP client only needs local storage.

mcp sidecar
$ docker run --rm --interactive --init --read-only --network none --cap-drop ALL --security-opt no-new-privileges=true --pids-limit 256 --tmpfs /tmp:rw,noexec,nosuid,size=64m --env-file lummenna.env --volume lummenna-store:/data/store --volume lummenna-models:/data/models lummenna-community:1.0.1rc4 lummenna-mcp
ready

08. Release boundaries

This release candidate is single-node, does not bundle an embedding model, and remains subject to API change. Community use is AGPL-3.0-or-later.