Sandbox Migrate To Next

Migrate stable → Sandbox SDK 1.0 preview (@next)

Perform the port. Follow the steps in order. Depth lives in docs—fetch the linked page when a step needs detail.

Human guide: Migrate (opens in a new tab) · 1.0 preview (opens in a new tab)

New projects should start on @next (sandbox-next), not this skill. Day-to-day stable work → sandbox-stable. Deprecated-API cleanup without moving to @next → 2026 deprecation guide (opens in a new tab) first if needed.

Existing apps should migrate when you can, so you are ready when 1.0 becomes the stable release. Do not force production cutover without the user agreeing.

Prefer installed @next types and the migrate doc over memory.

Workflow

1. Review hard rules and the replacement map
2. Audit the codebase; list hits and target shapes
3. Clarify with the user (cutover, bridge, Python image, unclear sites)
4. Upgrade package, image, and code
5. Validate

Stop after any step that needs a user decision.

Hard rules

• Worker package and container image must be the same @next line.
• Production cutover uses immediate container rollout. Stable and @next control protocols are incompatible both ways; gradual rollout leaves a broken mixed window. In-flight container work can stop.
• After cutover, await sandbox.exec(...) means process started, not command finished.
• Argv is as-is (no implicit shell). Shell syntax needs an explicit shell binary.
• Process handles have no stdin → terminals for interactive input.
• Observation timeout / AbortSignal cancel the wait only, not the process.
• No single retry loop for every error.
• Do not invent APIs (gitCheckout on core, process stdin, string-exec completion helper).
• Self-deployed bridge stays on stable (not part of the preview line yet).

Replacement map

Depth: Migrate (opens in a new tab) · after port, day-to-day → sandbox-next

Audit

rg 'SANDBOX_TRANSPORT|transport:|setTransport|enableDefaultSession|createSession|getSession|deleteSession|execStream\(|startProcess\(|killProcess\(|sandbox\.terminal\(|sessionId|gitCheckout\(|SandboxTransport|ExecutionSession'

Also: string exec(, cd then a later exec, bare createCodeContext / runCode on Sandbox.

Clarify (ask when needed)

• OK to cut production with --containers-rollout=immediate (live processes/terminals/streams may stop)?
• Self-deployed bridge? Leave on stable.
• Python interpreter → -python image variant?
• Call sites not covered by the map?

Upgrade

Package and image

npm install @cloudflare/sandbox@next

FROM cloudflare/sandbox:next
# Python: cloudflare/sandbox:next-python

Same prerelease tag on Worker and image when not on floating next.

Code by area

Apply replacements from the map. For each area, implement from the doc—not from stable habits:

Commands (shape):

// Before (stable)
const result = await sandbox.exec("npm test");

// After (@next)
const process = await sandbox.exec(["/bin/bash", "-lc", "npm test"]);
const result = await process.output({ encoding: "utf8" });

const server = await sandbox.exec(["/bin/bash", "-lc", "npm run dev"], {
  cwd: "/workspace/app",
});
await server.waitForPort(3000, { timeout: 60_000 });
await server.kill(); // numeric; default 15

Terminals (shape):

const terminal = await sandbox.createTerminal({ command: ["bash"], cwd: "/workspace" });
const t = await sandbox.getTerminal(terminal.id);
if (!t) return new Response("terminal gone", { status: 410 });
return t.connect(request, { cursor, cols, rows });

Interpreter (shape):

import { Sandbox as BaseSandbox } from "@cloudflare/sandbox";
import { withInterpreter } from "@cloudflare/sandbox/interpreter";

export class Sandbox extends BaseSandbox<Env> {
  interpreter = withInterpreter(this);
}

Git (shape):

const clone = await sandbox.exec(
  ["git", "clone", "--depth", "1", "--", repoUrl, "/workspace/repo"],
  { cwd: "/workspace" },
);
const result = await clone.output({ encoding: "utf8" });

Delete transport settings entirely. Remove session APIs. Isolate users with separate sandbox IDs.

Deploy cutover

Staging/branch first. Production is one deploy of matching Worker + image:

npx wrangler deploy --containers-rollout=immediate

Leave rollout_active_grace_period at default 0 (or set 0 if raised). After cutover, pre-deploy process/terminal IDs are invalid. Details: Migrate (opens in a new tab) · Container rollouts (opens in a new tab)

Validate

1. Lockfile + Dockerfile on the same @next line
2. Typecheck against @next
3. Smoke argv exec + output({ encoding: "utf8" })
4. Smoke long process / terminal / interpreter if used
5. Errors distinguished: unavailable / interrupted-RPC / stale / local wait
6. No live secrets in sandbox env
7. Grep again for removed APIs
8. Production used --containers-rollout=immediate

Then day-to-day work uses sandbox-next.

Red flags — stop and fix

• Mixing @next Worker with stable image (or reverse)
• Gradual container rollout for this cutover
• Treating await exec as command completion
• Assuming cd / exports persist across exec calls
• One retry wrapper for every error
• Inventing gitCheckout, process stdin, or undocumented APIs
• Keeping pre-cutover process/terminal IDs after deploy
• Forcing production cutover without user agreement
• Putting live secrets in setEnvVars / launch env