Deepline SDK

Beta: the SDK is usable, but its public surface is still evolving.

The Deepline SDK turns a repeatable GTM decision into a Play: TypeScript that Deepline can run, observe, and resume. You can start a Play from a coding agent, the CLI, an application, a webhook, or a schedule.

Mental model

Term	Meaning	Start here
Play	Your TypeScript business logic, executed by Deepline	Write Plays
Prebuilt Play	A Deepline-maintained Play with a published input and output contract	Prebuilt Plays
Tool	One provider-backed operation used inside a Play	SDK reference
Play run	One durable execution with status, output, and history	Call Plays
Dataset	A named collection of rows a Play reads or produces	Batch CSV enrichment
Runtime Sheet	Durable row storage backing a dataset during a run	SDK reference
Trigger	A declared event or schedule that starts a top-level Play run	Webhooks and schedules

A Play is code running in Deepline. A Monitor is different: it pushes an event from an external source into Deepline, which can then start a Play.

Choose a path

If you need to…	Go to
Prove one enrichment works before you write code	Quickstart
Find the closest outcome and its next guide	Examples
Write or change the GTM logic	Write Plays
Reuse a maintained workflow	Prebuilt Plays
Process a lead or account file	Batch CSV enrichment
Start work from an inbound event or a schedule	Webhooks and schedules
Start or observe work from a service	Call Plays
Integrate a production backend safely	Production Play integration

Where code runs

Location	Use	Typical work
Inside a Play	`definePlay(...)` and `ctx.*`	Persist rows, call tools, compose child Plays, and make recorded external requests
In an application or script	`Deepline.connect()` or `DeeplineClient`	Start a named Play, persist its run ID, poll, stream, or stop it
In a non-TypeScript system	The HTTP API	Call an already-saved or prebuilt Play from Python, a notebook, or a worker

Work safely

Validate a custom Play with deepline plays check before you deploy it or run provider-backed work. Then use a small, representative pilot and inspect the output before you scale. Write Plays owns the authoring check; Batch CSV enrichment owns the file-pilot procedure. For a maintained enrichment route, use the email waterfall.

Operate the lifecycle

Write and check a local Play, then run a representative pilot and review its output. Publish a stable version only before you attach a schedule, webhook, or other Trigger.

Get started

Build and automate

Common workflows

Data and integrations

Reference

Mental model

Choose a path

Where code runs

Work safely

Operate the lifecycle

Reference

SDK reference

Runtime API reference

​Mental model

​Choose a path

​Where code runs

​Work safely

​Operate the lifecycle

​Reference

SDK reference

Runtime API reference

Mental model

Choose a path

Where code runs

Work safely

Operate the lifecycle

Reference