Sandboxes

View as Markdown

A sandbox is a hardware-isolated virtual machine with its own kernel, filesystem, network stack, and allocated vCPU, RAM, and disk, with isolation enforced by the hypervisor.

Sandboxes run as Linux (default) or macOS virtual machines. Every sandbox moves through a lifecycle and can have several different states. While the sandbox is running, it exposes a toolbox for code, command, and process execution, sessions, filesystem and git operations, computer use, language server protocol support, and log streaming.

Use sandboxes to run code written by an agent, build and test software, drive desktop through screenshots and input events, or spin up thousands of identical environments for autonomous, evaluation, and reinforcement learning workflows.

Concept	Description
Sandbox	An isolated, dedicated virtual machine.
Image	The OCI image, Dockerfile build, or snapshot the sandbox boots from.
Resources	The vCPU, memory, and disk allocated to the sandbox.
Lifecycle	The states a sandbox can be in and the transitions between them.
Persistence	Deleting a sandbox discards its filesystem. Persistent macOS sandboxes can stop/start, pause/resume, fork.
Toolbox	A set of features for code, command, and process execution, sessions, filesystem and git operations, computer use, and language server protocol support.

Create sandboxes

Daytona provides methods to create sandboxes from the dashboard or programmatically.

Go to Daytona Sandboxes ↗
Click Create
Select Linux sandbox or macOS sandbox
Click Create

from daytona import Daytona

daytona = Daytona()
sandbox = daytona.v2.create()

import { Daytona } from '@daytona/sdk'

const daytona = new Daytona()
const sandbox = await daytona.v2.create()

daytona sandbox create

curl 'https://api.daytona.com/sandboxes' \
  --request POST \
  --header 'Content-Type: application/json' \
  --header 'Authorization: Bearer YOUR_API_KEY' \
  --data '{}'

Linux sandboxes

Linux sandboxes are Firecracker microVMs that run their own Linux kernel, and they are the default sandbox type. A Linux sandbox boots from an image that you import from a container registry, build from a Dockerfile, or capture from another sandbox. It can also start from a snapshot of another Linux sandbox. Each sandbox runs on an amd64 or arm64 CPU.

from daytona import Daytona

daytona = Daytona()
sandbox = daytona.v2.create()

import { Daytona } from '@daytona/sdk'

const daytona = new Daytona()
const sandbox = await daytona.v2.create()

daytona sandbox create

curl 'https://api.daytona.com/sandboxes' \
  --request POST \
  --header 'Content-Type: application/json' \
  --header 'Authorization: Bearer YOUR_API_KEY' \
  --data '{}'

macOS sandboxes

macOS sandboxes are virtual machines running macOS on Apple silicon Mac computers. A macOS sandbox starts from a macOS image. By default it gets half of the Mac’s cores, memory and sandbox disk on a split reservation, or all of them on a whole reservation; a create can ask for a smaller size.

from daytona import Daytona

daytona = Daytona()
sandbox = daytona.v2.create_macos("base-macos")

import { Daytona } from '@daytona/sdk'

const daytona = new Daytona()
const sandbox = await daytona.v2.createMacos('base-macos')

curl 'https://api.daytona.com/sandboxes/macos' \
  --request POST \
  --header 'Content-Type: application/json' \
  --header 'Authorization: Bearer YOUR_API_KEY' \
  --data '{"image": "base-macos"}'

Concept	Description
Sandbox	An isolated, dedicated virtual machine.
Image	The OCI image, Dockerfile build, or snapshot the sandbox boots from.
Resources	The vCPU, memory, and disk allocated to the sandbox.
Lifecycle	The states a sandbox can be in and the transitions between them.
Persistence	Deleting a sandbox discards its filesystem. Persistent macOS sandboxes can stop/start, pause/resume, fork.
Toolbox	A set of features for code, command, and process execution, sessions, filesystem and git operations, computer use, and language server protocol support.

Resources

Sandboxes have 0.125 vCPU, 1024MiB memory, and 512MiB disk by default. A sandbox may guarantee at most 16 vCPUs, 32GiB memory, and 256GiB disk.

Resource	Unit	Default	Minimum	Maximum
CPU	vCPU	`0.125`	`0.125`	`16`
Memory	MiB	`1024`	`512` (when set explicitly)	`32768`
Disk	MiB	`512`	`10240` (when set explicitly)	`262144`

from daytona import Daytona
from daytona.v2 import Resources

daytona = Daytona()
sandbox = daytona.v2.create(
    # memory and disk are GiB: 4 maps to memory_mib=4096, 20 to disk_mib=20480
    resources=Resources(cpu=2, memory=4, disk=20),
)

import { Daytona } from '@daytona/sdk'

const daytona = new Daytona()
const sandbox = await daytona.v2.create({
  // memory and disk are GiB: 4 maps to memory_mib=4096, 20 to disk_mib=20480
  resources: { cpu: 2, memory: 4, disk: 20 },
})

# --memory and --disk are MiB
daytona sandbox create --cpu 2 --memory 4096 --disk 20480

curl 'https://api.daytona.com/sandboxes' \
  --request POST \
  --header 'Content-Type: application/json' \
  --header 'Authorization: Bearer YOUR_API_KEY' \
  --data '{
  "cpu": 2,
  "memory_mib": 4096,
  "disk_mib": 20480
}'

Hardware virtualization

Server processors include instructions for running virtual machines directly on the hardware. A virtual machine monitor uses these instructions to create a guest with its own virtual processors, memory, and devices. The processor rejects any access from a guest to host memory or to another guest’s memory. Linux and macOS sandboxes run as virtual machines, each on the virtual machine monitor that its operating system needs.

Sandbox	Host	Runtime
Linux	Linux with KVM	Firecracker virtual machine
macOS	Mac with Apple silicon	Virtualization framework virtual machine

Linux exposes the processor’s virtualization through the Kernel-based Virtual Machine (KVM). Linux sandboxes run on Firecracker, a virtual machine monitor built to run many small, short-lived virtual machines on one host. Firecracker provides only the devices a server needs: virtual disks, a network card, and a serial console. Each Linux sandbox runs as one Firecracker process on a host, in its own network namespace, with its own virtual processors and memory.

macOS sandboxes run on Macs, because Apple’s license allows macOS virtual machines only on Apple hardware. Apple silicon includes hardware support for virtual machines, which macOS exposes through Apple’s Virtualization framework. Each macOS sandbox runs as its own virtual machine on its Mac, with its own virtual processors, memory, disk, and network card.

Lifecycle

Every sandbox moves through a lifecycle. A sandbox can have several different states. Each state reflects the status of your sandbox. A sandbox transitions between states in response to user actions, or automatically when one of its timeouts elapses.

See the lifecycle of Linux and macOS sandboxes.

Daytona Linux sandbox lifecycle state diagram with two states: Running and Deleted. The sandbox is ephemeral: sandbox.delete() moves Running to Deleted, the terminal state. .sandbox-diagram-fallback { margin: 2rem 0; }

Feature	Linux	macOS
Stop and start	✗	✓ persistent
Pause and resume	✗	✓ persistent
Fork	✗	✓ persistent
Delete	✓	✓
Filesystem persistence	✗	✓
Memory persistence	✗	✓ persistent
Memory snapshot	✓ warm template	✗
Filesystem snapshot	✓	✓ persistent
Spot (Preemption)	✓	✗
Network policy	✓	✓
Desktop	✓	✓ Screen Sharing

Operations

Daytona provides methods to manage existing sandboxes. Sandbox operations find sandboxes, read their current state and configuration, and update their metadata. They do not change the state of a sandbox or touch anything running inside it. To stop, pause, fork, or delete a sandbox, see lifecycle. To run commands or work with files inside a sandbox, use the toolbox.

List sandboxes

List sandboxes in the project, page by page. Unfiltered lists read the region-local database, so new sandboxes appear at once. To filter by name, state, labels, or resources, use search.

from daytona import Daytona

daytona = Daytona()
page = daytona.v2.list()
for sandbox in page.items:
    print(sandbox.id, sandbox.name, sandbox.state)

page.next_cursor is None on the last page.
Pass next_cursor back as cursor for the next page.
Passing any argument other than cursor and limit, for example name, status, or labels, sends the call to the search index instead.
The search index is eventually consistent.

import { Daytona } from '@daytona/sdk'

const daytona = new Daytona()
const page = await daytona.v2.list()
for (const sandbox of page.items) {
  console.log(sandbox.id, sandbox.name, sandbox.state)
}

page.nextCursor is unset on the last page.
Pass nextCursor back as cursor for the next page.
Passing any option other than cursor and limit, for example name, status, or labels, sends the call to the search index instead.
The search index is eventually consistent.

daytona sandbox list

Flag	Type	Description
`--limit`	`int`	Sets the page size. The default is `100`.
`--cursor`	`string`	Continues from a previous page.

The command prints rows from responding regions even when another region fails.

curl 'https://api.daytona.com/sandboxes' \
  --header 'Authorization: Bearer YOUR_API_KEY'

Field	Type	Description
`labels`	`string`	Filters by a JSON-encoded object such as `{"team":"research"}`; only sandboxes whose labels contain every pair are returned. A malformed value returns `400`.

Pass next_cursor from the response as cursor to read the next page.
The response includes deleted sandboxes until retention removes them.

Iterate over all sandboxes

Iterate over every sandbox in the project without handling page cursors. The iterator fetches the next page when the current one is exhausted and stops when the server returns no cursor.

from daytona import Daytona

daytona = Daytona()
for sandbox in daytona.v2.iter():
    print(sandbox.id, sandbox.name, sandbox.state)

import { Daytona } from '@daytona/sdk'

const daytona = new Daytona()
for await (const sandbox of daytona.v2.iter()) {
  console.log(sandbox.id, sandbox.name, sandbox.state)
}

# Follow every page until the server returns no cursor
daytona sandbox list --all

Search sandboxes

Search sandboxes across every region. Search reads a central index that is eventually consistent: a create, update, or delete becomes searchable about 2 to 3 seconds after it is committed, so a sandbox created a moment ago may be missing. To read your own writes, use list without filters or get. Filters combine with AND, and deleted sandboxes stay in the results until retention removes them unless you filter by state.

from daytona import Daytona

daytona = Daytona()

# Free text over name and image, or an exact sandbox id
page = daytona.v2.search(q="my-sandbox")
for sandbox in page.items:
    print(sandbox.id, sandbox.name, sandbox.state)

Parameter	Type	Description
`status`	`Sequence[str]`	API state names such as `running`, not the SDK names returned in `sandbox.state`.

The response has the following fields:

Field	Description
`next_cursor`	The cursor is tied to the filters and sort of its call.
`partial_regions`	Lists the regions that are backfilling. Results from these regions can be incomplete.

import { Daytona } from '@daytona/sdk'

const daytona = new Daytona()

// Free text over name and image, or an exact sandbox id
const page = await daytona.v2.search({ q: 'my-sandbox' })
for (const sandbox of page.items) {
  console.log(sandbox.id, sandbox.name, sandbox.state)
}

Parameter	Type	Description
`status`	`string \| readonly string[]`	API state names such as `running`, not the SDK names returned in `sandbox.state`.

The response has the following fields:

Field	Description
`nextCursor`	The cursor is tied to the filters and sort of its call.
`partialRegions`	Lists the regions that are backfilling. Results from these regions can be incomplete.

# Free text over name and image, or an exact sandbox id
daytona sandbox search my-sandbox

Flag	Type	Description
`--status`	`string`	Repeatable or comma-separated. Matches any listed value.
`--class`	`string`	Repeatable or comma-separated. Matches any listed value.

The query is optional.
Filter flags such as --status, --class, --name-prefix, and --labels select sandboxes on their own or narrow the query.

# Free text over name and image, or an exact sandbox id
curl 'https://api.daytona.com/sandboxes/search?q=my-sandbox' \
  --header 'Authorization: Bearer YOUR_API_KEY'

Field	Type	Description
`limit`	`integer`	Sets the page size, from `1` to `200`. The default is `100`.
`sort`	`string`	Sets the sort field. The default is `created_at`.
`order`	`string`	Sets the sort order. The default is `desc`.

A cursor is tied to the filters and sort of its request.

Get a sandbox

Get a sandbox by its ID. The response includes the current state, resources, labels, and lifecycle settings. Get reads the control plane directly, so it reflects changes immediately, unlike search.

from daytona import Daytona

daytona = Daytona()

# Fetch the sandbox from the control plane
sandbox = daytona.v2.get("{sandboxId}")
print(sandbox.state, sandbox.cpu, sandbox.memory, sandbox.disk)

The response has the following fields:

Field	Description
`memory`	Gives the memory in GiB, rounded up to a whole GiB. The API reports memory in MiB as `memory_mib`.
`disk`	Gives the disk in GiB, rounded up to a whole GiB. The API reports disk in MiB as `disk_mib`.

import { Daytona } from '@daytona/sdk'

const daytona = new Daytona()

// Fetch the sandbox from the control plane
const sandbox = await daytona.v2.get('{sandboxId}')
console.log(sandbox.state, sandbox.cpu, sandbox.memory, sandbox.disk)

The response has the following fields:

Field	Description
`memory`	Gives the memory in GiB, rounded up to a whole GiB. The API reports memory in MiB as `memory_mib`.
`disk`	Gives the disk in GiB, rounded up to a whole GiB. The API reports disk in MiB as `disk_mib`.

daytona sandbox get {sandboxId}

curl 'https://api.daytona.com/sandboxes/{sandboxId}' \
  --header 'Authorization: Bearer YOUR_API_KEY'

Find sandboxes by name

Find sandboxes by the name they were created with. Names are not unique, so the result is a list, newest first, and it is empty when no sandbox matches. The lookup reads the control plane directly, so a sandbox created a moment ago is found, unlike with search.

The name is stored as the name label. Replacing a sandbox’s labels without it removes the sandbox from name lookups.

from daytona import Daytona

daytona = Daytona()

# Reuse a named sandbox, or create it when none exists
found = daytona.v2.find_by_name("my-sandbox")
sandbox = found[0] if found else daytona.v2.create(name="my-sandbox")

Parameter	Type	Description
`name`	`str`	Required. Exact match. A blank name raises `BadRequestError` before the request is sent.
`region`	`str`	Sets the region to search. `"local"` or omitting it means the region of the API host.

find_by_name() leaves out deleted and deleting sandboxes.

import { Daytona } from '@daytona/sdk'

const daytona = new Daytona()

// Reuse a named sandbox, or create it when none exists
const found = await daytona.v2.findByName('my-sandbox')
const sandbox = found[0] ?? (await daytona.v2.create({ name: 'my-sandbox' }))

Parameter	Type	Description
`name`	`string`	Required. Exact match. A blank name throws `DaytonaError` before the request is sent.
`options.region`	`string`	Sets the region to search. `'local'` or omitting it means the region of the API host.

findByName() leaves out deleted and deleting sandboxes.

# Search by the name label: labels={"name":"my-sandbox"}
curl 'https://api.daytona.com/sandboxes/search?labels=%7B%22name%22%3A%22my-sandbox%22%7D' \
  --header 'Authorization: Bearer YOUR_API_KEY'

Unlike the SDK methods, the response can include deleting sandboxes. Use status to filter them out.
Search results can be briefly stale after a change.

Create an access token

Create a short-lived access token for one sandbox. The token reaches the sandbox toolbox, SSH, desktop, and private preview ports, and can stop, pause, or delete the sandbox, limited to the operations and ports you grant. Give it to a collaborator, a CI job, or a browser instead of a project API key.

The token lasts one hour by default and at most 24 hours. It cannot be refreshed, revoked, or used to create other credentials. Create a new token when it expires.

# Grant file system and SSH access for 15 minutes
token = sandbox.create_access_token(operations=["toolbox:fs", "ssh"], ttl_seconds=900)
print(token.access_token, token.expires_at)

Parameter	Type	Description
`operations`	`Sequence[str]`	Lists the operations the token grants. Each operation is one of `toolbox`, `toolbox:exec`, `toolbox:fs`, `toolbox:git`, `toolbox:lsp`, `desktop`, `ssh`, `sandbox:stop`, `sandbox:pause`, or `sandbox:delete`. The default is every operation the API key can grant. `[]` grants preview ports only. Granting `sandbox:delete` requires the `delete:sandboxes` permission.
`ports`	`Sequence[int]`	Sets the private preview ports that the token can reach. `[]` grants none. The default is all ports.
`ttl_seconds`	`int`	Sets the token lifetime in seconds, at most `86400`. The default is `3600`.

// Grant file system and SSH access for 15 minutes
const token = await sandbox.createAccessToken({ operations: ['toolbox:fs', 'ssh'], ttlSeconds: 900 })
console.log(token.accessToken, token.expiresAt)

Parameter	Type	Description
`operations`	`string[]`	Lists the operations the token grants. Each operation is one of `toolbox`, `toolbox:exec`, `toolbox:fs`, `toolbox:git`, `toolbox:lsp`, `desktop`, `ssh`, `sandbox:stop`, `sandbox:pause`, or `sandbox:delete`. The default is every operation the API key can grant. `[]` grants preview ports only. Granting `sandbox:delete` requires the `delete:sandboxes` permission.
`ports`	`number[]`	Sets the private preview ports that the token can reach. `[]` grants none. The default is all ports.
`ttlSeconds`	`number`	Sets the token lifetime in seconds, at most `86400`. The default is `3600`.

# Grant file system and SSH access for 15 minutes
curl 'https://api.daytona.com/sandboxes/{sandboxId}/access-token' \
  --request POST \
  --header 'Content-Type: application/json' \
  --header 'Authorization: Bearer YOUR_API_KEY' \
  --data '{"operations": ["toolbox:fs", "ssh"], "ttl_seconds": 900}'

Field	Type	Description
`operations`	`string[]`	Sets the operations that the token grants: any of `toolbox`, `toolbox:exec`, `toolbox:fs`, `toolbox:git`, `toolbox:lsp`, `desktop`, `ssh`, `sandbox:stop`, `sandbox:pause`, and `sandbox:delete`. `[]` grants preview ports only. The default is every operation that the API key can grant.
`ports`	`integer[]`	Sets the private preview ports that the token can reach, from `1` to `65535`, at most 128. `[]` grants none. The default is all ports.
`ttl_seconds`	`integer`	Sets the token lifetime in seconds, at most `86400`. The default is `3600`.

The response has the following fields:

Field	Description
`access_token`	The access token.
`token_type`	The token type is `Bearer`.
`expires_at`	The time when the token expires.
`operations`	The operations that the token grants.
`ports`	Lists the private preview ports the token can reach. The value is `null` when every port is granted.

The request requires the write:sandboxes permission.
Authenticate with an API key.
A sandbox access token cannot create another token.

Update labels

Update a sandbox’s labels. Labels are key-value strings for organizing sandboxes, and you can filter by them when you list or search. An update replaces the full label set; omitted labels are removed. A sandbox holds at most 100 labels, each key and value at most 512 bytes.

from daytona import Daytona

daytona = Daytona()
sandbox = daytona.v2.create()

# Replace the full label set
labels = sandbox.set_labels(labels={"team": "research"})
print(labels)

Parameter	Type	Description
`labels`	`Mapping[str, str]`	Required. Keyword only. `{}` removes all labels.

set_labels() updates sandbox.labels.

import { Daytona } from '@daytona/sdk'

const daytona = new Daytona()
const sandbox = await daytona.v2.create()

// Replace the full label set
const labels = await sandbox.setLabels({ team: 'research' })
console.log(labels)

Parameter	Type	Description
`labels`	`Record<string, string>`	Required. `{}` removes all labels.

setLabels() updates sandbox.labels.

# Replace the full label set
daytona sandbox update {sandboxId} --label team=research

Flag	Type	Description
`--label`	`string`	Repeatable, one `key=value` pair per flag.
`--clear-labels`	`bool`	Removes all labels.

# Replace the full label set
curl 'https://api.daytona.com/sandboxes/{sandboxId}/labels' \
  --request PUT \
  --header 'Content-Type: application/json' \
  --header 'Authorization: Bearer YOUR_API_KEY' \
  --data '{ "labels": { "team": "research" } }'

Field	Type	Description
`labels`	`object`	Replaces the full label set. `{}` removes all labels. A sandbox holds at most 100 labels. Each key and value is at most 512 bytes.

Logs

Read the serial console log of a sandbox. For a running sandbox the log is read live from its runner; for a sandbox in any other state the persisted log is returned. macOS sandboxes do not serve serial console or entrypoint logs.

# Serial console log of the microVM
daytona sandbox logs {sandboxId}

Flag	Type	Description
`--follow`	`bool`	Polls for new log output until interrupted.
`--entrypoint`	`bool`	Reads the entrypoint logs instead of the serial console.

# Serial console log of the microVM
curl 'https://api.daytona.com/sandboxes/{sandboxId}/console-logs' \
  --header 'Authorization: Bearer YOUR_API_KEY'

Entrypoint logs

Read the log of the sandbox entrypoint process as text. For a running sandbox the log is read live; for a sandbox in any other state the persisted log is returned. To page through the log or follow it, read the entrypoint process through the toolbox.

print(sandbox.entrypoint_logs())

console.log(await sandbox.entrypointLogs())

daytona sandbox logs {sandboxId} --entrypoint

curl 'https://api.daytona.com/sandboxes/{sandboxId}/logs' \
  --header 'Authorization: Bearer YOUR_API_KEY'

Build logs

Read the output of the image build behind a sandbox. A sandbox has a build log when its image is built from a Dockerfile, including a build that failed before the sandbox was placed; a sandbox created from a registry image has none. When the sandbox reuses an image built earlier, the log is the output of that original build.

Following the log streams output while the build runs and ends when the build finishes. Stopping the reader does not stop the build. To read the log of an image without a sandbox, see image build logs.

from daytona import Daytona
from daytona.v2 import ImageBuilder

daytona = Daytona()
sandbox = daytona.v2.create_from_build(ImageBuilder.from_dockerfile("FROM python:3.12-slim"))

# Output retained so far, as text
print(sandbox.build_logs())

# Stream output until the build finishes
result = sandbox.follow_build_logs(on_logs=lambda chunk: print(chunk, end=""))
print(result.status, result.reason)

Parameter	Type	Description
`on_logs`	`Callable[[str], None]`	Receives each chunk of decoded output. This parameter is required.
`cursor`	`str`	Resumes after the output a previous follow returned in `cursor`.
`timeout`	`float`	Sets the follow limit in seconds. By default, there is no limit.

The response has the following fields:

Field	Description
`status`	The status of the build.
`reason`	The value is `terminal`, `deadline`, `not_applicable`, or `resource_deleted`.
`output_complete`	Whether the build output is complete.
`cursor`	Pass this value as `cursor` to resume after this output.

build_logs() raises NotFoundError when the sandbox has no image build.
follow_build_logs() raises BuildLogInterruptedError when the sandbox has no image build.
BuildLogInterruptedError carries the HTTP status and the last cursor.

import { Daytona } from '@daytona/sdk'
import { ImageBuilder } from '@daytona/sdk/v2'

const daytona = new Daytona()
const sandbox = await daytona.v2.createFromBuild(ImageBuilder.fromDockerfile('FROM python:3.12-slim'))

// Output retained so far, as text
console.log(await sandbox.buildLogs())

// Stream output until the build finishes
const result = await sandbox.followBuildLogs({ onLogs: (chunk) => { process.stdout.write(chunk) } })
console.log(result.status, result.reason)

Parameter	Type	Description
`onLogs`	`(chunk: string) => unknown`	Required. Receives each chunk of decoded output.
`cursor`	`string`	Resumes after the output a previous follow returned in `cursor`.
`timeout`	`number`	Sets the follow limit in milliseconds, unlike the seconds that the Python SDK uses. By default, there is no limit.

The response has the following fields:

Field	Description
`status`	The status of the build.
`reason`	The value is `terminal`, `deadline`, `not_applicable`, or `resource_deleted`.
`outputComplete`	Whether the build output is complete.
`cursor`	Pass this value as `cursor` to resume after this output.

buildLogs() throws NotFoundError when the sandbox has no image build.
followBuildLogs() throws BuildLogInterruptedError when the sandbox has no image build.
BuildLogInterruptedError carries the HTTP status and the last cursor.

BUILD_LOGS_URL=$(curl 'https://api.daytona.com/sandboxes/{sandboxId}/build-logs-url' \
  --header 'Authorization: Bearer YOUR_API_KEY' | jq -r '.url')

# Output retained so far, as text
curl "$BUILD_LOGS_URL?follow=false&format=text" \
  --header 'Authorization: Bearer YOUR_API_KEY'

# Stream output until the build finishes
curl --no-buffer "$BUILD_LOGS_URL?follow=true&format=sse" \
  --header 'Authorization: Bearer YOUR_API_KEY' \
  --header 'Accept: text/event-stream'

The first request returns 404 when the sandbox has no image build.
A request returns 503 when the build log is unavailable, including a build removed by garbage collection.