Custom Python snapshot

Bake a python:3.12 image with pandas / numpy / requests preinstalled.

A fully custom snapshot using defineSnapshot. No preset, no convention — you describe what you want.

import { buildSnapshot, defineSnapshot } from "@beamhop/lightbox";

const pythonDataSnapshot = defineSnapshot({
  name: "py-data",
  image: "python:3.12",
  resources: { cpus: 2, memory: "1G" },
  workdir: "/work",
  setup: [
    "pip install --no-cache-dir numpy pandas requests",
  ],
  labels: { team: "data" },
});

await buildSnapshot(pythonDataSnapshot);
console.log(`Snapshot "${pythonDataSnapshot.name}" ready.`);

Once built, run code in an ephemeral sandbox with runInSandbox — it boots, runs your callback, and tears down even if you throw:

import { runInSandbox } from "@beamhop/lightbox";

const out = await runInSandbox(
  { snapshot: "py-data", name: "analysis-1" },
  (sb) => sb.shell(`python -c "import pandas as pd; print(pd.__version__)"`),
);
console.log(out.stdout());