Louhin CLI

Command-line tool for local development of Louhin code modules: browser-SSO login, pull modules and data to disk, edit them in your favourite editor with full debug support, and push changes back.

Install

npm install -g @louhin/cli

Requires Node.js 20+. For running modules locally you also need Python 3.10 or newer.

First-time setup

louhin config set server https://example.com
louhin login

The login command opens your browser, you sign in with the same credentials you use on the web UI, and the CLI receives a token on a loopback URL. Tokens are stored in ~/.config/louhin/credentials.json (mode 0600). The server URL is remembered so later commands don't need --server.

Typical workflow

louhin module list
louhin module pull 12345 --ide=vscode
cd My_Module
python -m venv .venv && .venv/Scripts/activate && pip install -r requirements.txt
louhin data pull 67890        # if your script reads input data
# edit files/main.py in your editor; set breakpoints, debug
louhin run
louhin module push            # ship changes back to the server

--ide=vscode, --ide=pycharm, and --ide=intellij are supported. The CLI drops a .vscode/launch.json or .idea/ scaffolding so the Louhin: Run configuration is ready to use on first open.

Commands

louhin login [--server <url>]
louhin logout [--server <url>]

louhin config                          # show current config
louhin config set <key> <value>
louhin config get <key>

louhin module list
louhin module pull <id> [dir]          # also accepts --ide=<vscode|pycharm|intellij>
louhin module push [dir]
louhin run [--entry <file>] [--param KEY=VALUE ...]

louhin data list [--project <id>]
louhin data pull <id> [name]           # writes data/<name>.json (default: in1, in2, ...)
                                       # --all to pull the full dataset
                                       # --anonymize for synthetic values

Pulled module layout

My_Module/
├── module.json           server URL + module id, used by push
├── pyproject.toml        Python project marker
├── requirements.txt      default analytics stack; edit as needed
├── README.md             quick reference for this module
├── files/                user-editable source code (Sources Root)
│   └── main.py
├── data/                 louhin+json fixtures
│   ├── in1.json          inputs your script reads as `in1`, `in2`, ...
│   └── out/              louhin run writes outputs here
└── .louhin/              CLI state (gitignored)
    └── runner.py         local runner shim

louhin run binds the same globals as the server-side runtime (inputs, in1, in2, ..., params, state, render, ctx) so the same code works identically locally and on the server.

Config files

• ~/.config/louhin/config.json -- default server, preferences
• ~/.config/louhin/credentials.json -- refresh + access tokens per server (mode 0600)

Override the server per command with --server <url> or the LOUHIN_SERVER env var.

License

MIT