Documentation
Up and remembering in minutes.
The full reference — README, integration guides, whitepaper, security policy — lives in the GitHub repository. This page covers the fast path.
1 · Install
Python 3.10+ on macOS, Linux, or Windows. From source:
git clone https://github.com/QuantmindSSI/lummenna.git
cd lummenna
pip install -e . # lean runtime — no torch/CUDA
Or install the packaged wheel from the free Community download:
pip install lummenna-1.1.0-py3-none-any.whl
2 · Initialize
lummenna init --device generic
# Store and retrieve from the CLI
lummenna memory store "We chose SQLite for the cache layer" --room decisions
lummenna memory retrieve "cache layer"
Lummenna Pro adds the REST server + dashboard:
export LUMMENNA_LICENSE_KEY='LUMMENNA-…'
lummenna serve
# Dashboard at http://localhost:8848/dashboard
# API docs at http://localhost:8848/docs
First run needs an embedding model — the lean path is a prebuilt bundle via LUMMENNA_PREBUILT_MODEL_URL; see the README for the export toolchain alternative.
3 · Attach it to your agent (MCP)
Lummenna ships an MCP server over stdio (requires the [mcp] extra: pip install 'lummenna[mcp]'). Ready-made configs for ten clients are in the repo's integrations/ directory: Claude Desktop, OpenAI Codex CLI, GitHub Copilot, Cline, Zed, opencode, Continue, Goose, Aider, and Devin.
Claude Desktop (claude_desktop_config.json):
{
"mcpServers": {
"lummenna": {
"command": "python",
"args": ["-m", "lummenna.integrations.mcp_server"],
"env": {
"LUMMENNA_DEVICE": "generic",
"LUMMENNA_STORE_PATH": "${userHome}/.lummenna/store",
"LUMMENNA_MODEL_PATH": "${userHome}/.lummenna/models"
}
}
}
}
The server exposes seven tools: lummenna_search, lummenna_store, lummenna_assemble, lummenna_turn, lummenna_feedback, lummenna_status, and lummenna_dashboard.
4 · Or use it from code
LangChain (LummennaChatMemory) and LangGraph (LummennaCheckpointSaver) adapters are built in, and the FastAPI server publishes an OpenAPI schema at /openapi.json for anything that speaks REST.
5 · Going Pro
Activation is one environment variable — no code changes:
export LUMMENNA_LICENSE_KEY='LUMMENNA-…'
lummenna license # verify: edition, seats, expiry — fully offline
See pricing for what Pro unlocks.
Further reading
README & source · INTEGRATIONS.md · SECURITY.md · ROADMAP.md