Skip to content

Dagy / Developer documentation

From your first flow
to a working integration.

Dagy lets you define data workflows in Python or on a visual canvas, deploy them to an environment, and start and inspect runs through a CLI or REST API.

Choose your starting point

Your goalStart hereWhat you will have
Understand DagyFlows, tasks, and runsA vocabulary for building and operating a workflow
Try it on your machinePython quickstartA working flow and a local run you can inspect
Integrate an applicationConnect to DagyCredentials, a deployed flow, and an API-triggered run
Compose configured nodesVisual flow builderA validated graph and an environment deployment
Find a request or responseAPI referenceEndpoint guides and a downloadable OpenAPI contract

What you can build

Use tasks to call your application's services, transform records, check data quality, write to a destination, or run an AI model. Pass task results to other tasks to declare dependencies. Start runs on demand, on a schedule, or through an inbound webhook. Inspect task states and logs, then retry or backfill when necessary.

The node catalog describes the provided ingestion, transformation, AI, vector database, and notification nodes. You can also use your own Python libraries. A catalog entry describes a capability and its configuration; credentials, installed packages, network access, and documented runtime limitations still determine whether it will execute successfully.

Understand availability before deploying

These docs describe the implementation in SDK version 0.1.0. Individual hosted capabilities depend on the deployment you use. Cloud-account setup currently supports registration and verification; it does not yet supply credentials to workflow tasks. Saved connector records also do not automatically resolve credentials for node executions. Integration guides explain the working alternatives.

Review limits and constraints and the production guide before moving beyond a development environment. The release information records compatibility notes without implying features have shipped to every hosted deployment.

Explore the documentation