Install once.
Use it naturally.
LOLM runs on your computer. The website does not execute prompts, touch files, or store provider credentials.
Install the command
npm install -g lolm-cli lolm --version
Node.js 20 or newer is required.
Connect your model
lolm setup lolm doctor
The guided setup works in Windows PowerShell, macOS, and Linux. Choose a built-in provider, Ollama, or a custom OpenAI-compatible endpoint. Your credentials stay in your local setup.
Other provider options
lolm setup openai lolm setup anthropic lolm setup ollama
You can also use your provider's normal environment variable. Set it using your shell's syntax; PowerShell and bash use different commands.
Open the terminal UI
lolm lolm tools lolm doctor
Keep talking in the same session. LOLM records durable local runs and can use typed terminal, filesystem, Git, GitHub, Cloudflare, browser, and computer tools. Say try again after a failed task, or use lolm run for one task in your project folder. Start with a small change, then inspect the files and checks:
lolm run "inspect this project and explain how to run its tests" lolm "answer this question…" --once lolm "make me a PDF and put it on my Desktop" lolm "code a polished HTML page here" --cwd ./project lolm "update yourself"
Advanced: enable the trained local controller
Optional NFET setup
The npm command works by itself. To use the trained NFET graft, clone the source and point LOLM at it:
git clone https://github.com/TheArtOfSound/LOLM.git cd LOLM python3 -m venv .venv .venv/bin/pip install -r requirements-agent.txt export LOLM_HOME="$PWD" lolm nfet status lolm nfet test "verified result text"
LOLM reports NFET unavailable when the model or checkpoint is missing. It never substitutes fake telemetry.
Continue the same workflow
Have a Flows plan? Copy its current instruction into LOLM, work in your project folder, then return the diff and test output to the same Flows run. Your local provider setup is separate from your Qira account.
Return to your Flows evidence → · See how Qira fits together →
Keep it current
lolm update --check lolm update # or simply lolm "update yourself"
Open source with a commercial path
LOLM is AGPL-3.0-or-later. Organizations that need to embed or host it without AGPL source-sharing obligations can request commercial terms.