@rtemis/launch
v0.1.1
Published
Start rtemis.server and ollama locally and open live.rtemis.org pre-authenticated
Downloads
47
Readme
@rtemis/launch
Start a local rtemis.server and open live.rtemis.org already authenticated.
Install & run
Run once without installing:
npx @rtemis/launchOr install globally and use the rtemislive command:
npm install -g @rtemis/launch
rtemisliveTo upgrade later: npm install -g @rtemis/launch@latest.
What it does:
- Generates a fresh token (
crypto.randomUUID()). - Starts the R server:
Rscript -e 'rtemis.server::serve(port=5757, token="…")'. - Starts
ollamafor the Chat panel, if installed (skipped if already running), withOLLAMA_ORIGINS=https://live.rtemis.orgso the app is allowed to connect to it from the browser. - Waits for the server to accept connections.
- Opens
https://live.rtemis.org/#rtemis=<payload>in your default browser. The app reads the token from the URL hash (never sent to any server), connects to your local server, and strips the token from the address bar.
Press Ctrl-C to stop the server (and ollama, if we started it).
Options
| Flag / env | Default | Meaning |
| ---------------------- | ------- | ----------------------------- |
| --port <n> | 5757 | Port for rtemis.server |
| RTEMIS_SERVER_PORT | 5757 | Same, via environment |
Requirements
- Node.js ≥ 18
- R with the
rtemis.serverpackage installed (RscriptonPATH) ollama(optional — only for local LLM Chat)
Why the token is safe in the URL
The token travels in the URL fragment (#rtemis=…), which browsers keep
client-side and never include in HTTP requests — so it never reaches the CDN
that serves the static site, nor any access log. On load the app decodes it,
connects to 127.0.0.1, and calls history.replaceState to remove it from the
address bar and history.
