Skip to main content

Tilebox Logo

Documentation | Console | Example Gallery

Tilebox Workflows

Tilebox Workflows, or the Tilebox workflow orchestrator is a parallel processing engine that allows an intuitive creation of dynamic tasks that can be parallelized out of the box and executed across compute environments or on-premise as well as in auto-scaling clusters in public clouds.

Quickstart

Install using pip:

pip install tilebox-workflows

For interactive job progress in Jupyter notebooks, install tilebox-workflows[notebook]. Without this extra, jobs use their plain-text representation. For S3 client type information, install boto3-stubs[s3] in your development environment.

Create a task:

from tilebox.workflows import Task


class MyFirstTask(Task):
    def execute(self):
        print("Hello World from my first Tilebox task!")

Submit a job

from tilebox.workflows import Client

# create your API key at https://console.tilebox.com
client = Client(token="YOUR_TILEBOX_API_KEY")

jobs = client.jobs()
jobs.submit("my-very-first-job", MyFirstTask())

And run it:

runner = client.runner(tasks=[MyFirstTask])
runner.run_all()

Workflow-scoped tasks

To submit tasks to a managed workflow, pass its remote reference or a Workflow object:

jobs.submit("scoped-job", MyFirstTask(), workflow="my-workflow")

workflow = client.workflows().find("my-workflow")
jobs.submit("another-job", MyFirstTask(), workflow=workflow)
jobs.query(workflows=[workflow, "tilebox:shared-workflow"])

The workflow's immutable UUID scopes each root task. A failed reference lookup raises an error without submitting. Omitting workflow keeps roots unscoped, including jobs submitted from inside another task. Existing direct runners continue to register and execute unscoped tasks.

Task metadata, runner registries, and release contents use UnboundTaskIdentifier (name and version). Submissions and executions use TaskIdentifier, which also carries the optional workflow ID. Use .bind(workflow_id) to create an execution identifier and .unbind() to recover its implementation identifier.

Children inherit the executing parent's scope. Execution through a legacy unscoped alias therefore produces unscoped children. The runner handles inheritance internally; context.submit_subtask() and context.submit_subtasks() do not expose workflow selection. Cross-workflow and explicitly unscoped child overrides are deferred until the CLI can distinguish scope-aware workers from workers that need its compatibility adapter.

Both create_cron_automation() and create_storage_event_automation() accept the same workflow parameter as jobs.submit(). Workflow objects expose id, reference, project_slug, and legacy_task_aliases; .slug remains a read-only alias for the remote reference. When constructing a Workflow directly, use reference= instead of slug=. The project slug is not a unique remote selector. Workflows sort by reference, name, and ID, with missing IDs first; other metadata does not affect ordering.

client.workflows().create(name, summary) always creates a scoped-only workflow; creating legacy workflows is not supported. Existing workflows still expose their alias policy. A None policy in a response means the server did not report that field, rather than reporting False.

Concurrent worker execution

A worker runtime can execute multiple tasks concurrently in one Python process. Each execution receives a newly deserialized task instance and its own ExecutionContext, including task-local subtask and progress state. The RunnerContext, configured JobCache, and any class or module state are process-level resources shared by those executions.

Custom runner contexts, caches, and shared task state must therefore support concurrent access from multiple threads. Asynchronous task executions may also run on different event loops. Configure and register these resources before the worker starts; do not mutate runner configuration while tasks are executing. Compound cache operations are not atomic unless the cache implementation explicitly provides that guarantee.

Concurrency in one runtime avoids repeated process initialization and allows overlapping I/O or native code that releases Python's GIL. CPU-bound Python code still needs multiple runtime processes for parallel execution.

Reading automation objects

Storage-event tasks can read objects from Amazon S3, Google Cloud Storage, Azure Blob Storage, or the local filesystem:

content = self.trigger.storage.read(self.trigger.location)

The method returns bytes. Cloud reads use the credentials configured for the runner.

On macOS, GCS or Azure CLI authentication can emit gRPC fork diagnostics even when a read succeeds. These messages alone do not indicate a failed read; do not downgrade gRPC to hide them.

Documentation

Check out the Tilebox Workflows documentation for more information.

License

Distributed under the MIT License (The MIT License).

Metadata

Release files for tilebox-workflows 0.68.0

For a detailed explanation of source distributions (sdists) and built distributions (wheels), please see the package formats documentation.

Source distribution (sdist)

Source distribution for tilebox-workflows 0.68.0
File Size Uploaded Provenance Actions
tilebox_workflows-0.68.0.tar.gz 145.5 kB With provenance Details for tilebox_workflows-0.68.0.tar.gz

Built distribution (wheel)

Table of built distributions (wheels) for tilebox-workflows 0.68.0
File Interpreter ABI Platform Actions
tilebox_workflows-0.68.0-py3-none-any.whl Python 3 none any Details for tilebox_workflows-0.68.0-py3-none-any.whl

Total release size: 327.4 kB

Release files / tilebox_workflows-0.68.0.tar.gz

Download URL tilebox_workflows-0.68.0.tar.gz
Size 145.5 kB
Tags Source
SHA-256 checksum
How to use checksums
b0d6a94682639f445e3711a98d0e606720b9c149760aee0a55d99e61a925816f
BLAKE2b-256 checksum
How to use checksums
1863e6ccdcc9579d8af18628d72cc5ec14dcb1c4019b2f221b7b34230948c29d
Upload date
Uploaded using Trusted Publishing?
What is trusted publishing?
Yes
Uploaded via twine/7.0.0 CPython/3.13.16

Provenance

Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.

PyPI Publish Attestation

PyPI verified that this artifact, at this checksum, originated from the publisher listed below.

Signed by GitHub Actions, verified by PyPI on Oct 30, 2026.

Transparency log

Release files / tilebox_workflows-0.68.0-py3-none-any.whl

Download URL tilebox_workflows-0.68.0-py3-none-any.whl
Size 182.0 kB
Tags Python 3
SHA-256 checksum
How to use checksums
15b66feb0e043ec46417d6264c41f45597189a3750cd6644058945e7d084e32b
BLAKE2b-256 checksum
How to use checksums
343592fab9838303ca9ff611fb29367678261789fde2094572c7555c8d6db75c
Upload date
Uploaded using Trusted Publishing?
What is trusted publishing?
Yes
Uploaded via twine/7.0.0 CPython/3.13.16

Provenance

Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.

PyPI Publish Attestation

PyPI verified that this artifact, at this checksum, originated from the publisher listed below.

Signed by GitHub Actions, verified by PyPI on Oct 30, 2026.

Transparency log

Security for tilebox-workflows 0.68.0

Provenance

Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.

Full provenance

Both files in this release have been uploaded with provenance metadata.

Select from 2 files

Attestation for tilebox_workflows-0.68.0.tar.gz

PyPI Publish Attestation

PyPI verified that this artifact, at this checksum, originated from the publisher listed below.

Signed by GitHub Actions, verified by PyPI on Oct 30, 2026.

Transparency log

A note on package security

PyPI is a repository for distributing packages, not a service for auditing them. While third-party security researchers and our staff do their best to detect and remove malicious packages, PyPI cannot guarantee the quality, security or suitability of any given package.

To learn more about protecting yourself from malicious actors, visit our package security documentation.

Release history Release notifications | RSS feed

This release

0.68.0 This release

2 release files Full provenance

Anthropic, PBC Visionary sponsor Bloomberg Visionary sponsor Hudson River Trading Visionary sponsor Meta Visionary sponsor NVIDIA Visionary sponsor Microsoft Sustainability sponsor Depot Continuous Integration AWS Cloud computing and Security Sponsor Datadog Monitoring Fastly CDN Google Download Analytics Sentry Error logging StatusPage Status page