Skip to content

Changelog

New in Prefect: features, improvements, and fixes. For the complete record of every open source release, see the release notes.

Cloud

Admin dashboards and archived flows

Account administrators get dashboards that report on every workspace at once, and flows you no longer use can be archived without losing their run history.

Admin dashboards

Until now, finding out how much your account ran, what failed, or who was using Prefect meant opening each workspace in turn. Account administrators can now build dashboards that answer those questions for every workspace on one page.

Every tile on a dashboard reads from all your workspaces at once. A single tile can show this week's failure rate across the whole account, set it against last week's, and change color when it crosses a threshold you choose. Other tiles track run hours, SLA violations, plan usage, members, object inventories and security.

Start from a template. The administrator template shows plan limits, members and idle workspaces. The operator template shows run volume, failure rate, run hours and SLA violations. If neither fits, build your own dashboard from scratch.

Configure each tile. Add color-coded thresholds, compare a value with the previous period, and pick whether run time or lateness shows as a mean, median, p90, p99 or maximum.

The account administrator dashboard templateThe account administrator dashboard template

Archive flows instead of deleting them

A workspace that has run for a year or two piles up flows. Renamed flows leave their old names behind, retired pipelines clutter the Flows page, and until now the only way to clear one out was to delete it, along with its run history and logs.

Now you can archive it. An archived flow leaves the Flows page, and its deployments leave the Deployments page. Its run history and logs stay, subject to your plan's run retention, and you can unarchive it at any time.

Archive one flow or many. Archive a flow from its menu on the Flows page, from the flow page itself or from the command palette, or select several flows and archive or unarchive them together. You need the manage_flows permission.

Find archived flows when you need them. Search, the command palette and flow pickers still list archived flows, marked Archived. The ⋮ menu beside the tabs on the Flows and Deployments pages switches between Unarchived, All and Archived, and a search that matches archived flows offers a link to show them.

Pause schedules as you archive. If a flow's deployments have active schedules, the archive dialog lists them and can pause them and archive the flow in one step. It also warns you when an automation runs one of the flow's deployments. Unarchiving leaves the schedules paused until you turn them back on.

Let Prefect archive dormant flows. Two workspace policies under Workspace Settings → Policies do the cleanup for you. Both are off by default, and an account administrator first has to allow workspace policies under Account Settings → Controls. Dormant flow archival archives flows that haven't run in a set number of days. The default is 90, and the window never exceeds your plan's run retention. It skips flows with an active schedule or with a deployment that an enabled automation runs, and it archives at most 50 flows per workspace per day, so a large first cleanup takes a few days. Archived flow restoration unarchives a flow the moment it gets a new run, which catches flows that a script or another orchestrator still runs. Turn it on whenever you turn on dormant flow archival.

Choose what archiving means in your workspace. With restoration on, archiving hides a flow until something runs it again. With restoration off, archiving retires the flow, and the UI won't let anyone start a run of it. Automations, API calls and re-enabled schedules can still run an archived flow, so pause or delete its deployments if you want it fully stopped.

Automate it. Automations get Archive a flow and Unarchive a flow actions, plus three new events to trigger on: prefect.flow.archived, prefect.flow.unarchived and prefect.flow.dormant. The API can archive or unarchive one flow or many and filter flows by archived. To read automations that use the new actions from the CLI or Python client, you need prefect 3.9.0 or later, which requires Python 3.11 or later.

Deployments can't be archived on their own; a deployment is hidden only when its flow is archived. Archiving isn't available in the open-source Prefect server.

To set it up, read the guide to archiving flows.

Cloud

Artifacts get a home in Prefect Cloud

A top-level Artifacts page in the sidebar and command palette, with redesigned collection and detail views, support for every SDK artifact type, and inline version rendering.

Artifacts now have a home in Prefect Cloud. They live in the sidebar and the command palette, so you can navigate or search straight to a full view of everything your workflows have produced, in one place. Alongside that, the collection and detail pages have been rebuilt to keep pace with everything the SDK can now attach to a run.

The collection view shows a preview of each artifact, its type, and when it was created, along with type-specific metadata like how far a progress artifact has advanced or how many rows a table holds. Every artifact type the SDK supports is represented, including progress and result artifacts.

The detail view is streamlined around the thing you came to see. Each version renders inline, with the raw form a click away on the same page, so there's no secondary page to hunt through. When you're looking at an older version, a badge tells you so and links you to the latest. You can still jump to the flow run that generated an artifact, and from a run, deep links open the exact artifact version it produced.

Progress artifacts also gained a sense of staleness. If a run finishes without its progress artifact ever reaching 100%, we mark that version stale rather than leaving the last reported number sitting there as if it were still live.

Cloud

Launch Week: dbt orchestration and infrastructure debugging in Cloud

Prefect Cloud's spring launch week brought first-class dbt orchestration, infrastructure debugging in the run page, managed automations, config plugins, and per-workspace rate limits.

Prefect dbt orchestrator. Run dbt projects with state-aware caching that skips work already fresh, plus per-model retries and concurrency management.
Infrastructure debugging in the run page. See memory pressure, pod logs, and eviction details for a run without leaving Prefect.
Managed automations. Streamlined automation management for enterprise workflows.
Config plugins. A plugin system for customizing deployment configuration.
Per-workspace rate limits. Set API rate-limit controls at the workspace level.

Read the full story on the launch page.

Open source
Cloud

Assets: track what your workflows produce

Replace @task with @materialize and Prefect tracks the tables, models, and files your workflows produce, and how they all connect.

When a dashboard throws garbage numbers at quarter-end, someone has to follow the breadcrumbs back through tasks, tables, and stray CSVs. That is asset lineage in the real world, and Prefect now tracks it without asking you to contort your workflow around asset definitions. Tell Prefect what a function produces:

assets.py
@materialize("s3://bucket/important-data.csv")
def write_data():
    # writes s3://bucket/important-data.csv
    ...

Prefect builds the lineage graph from there: what each workflow produces, when it last ran, and how your outputs depend on each other.

Read the announcement post for the full story, or start with the assets documentation.