Pipeline Editor

The pipeline editor is where you assemble, configure, and submit pipelines. It is built around a menu bar at the top, an infinite canvas in the middle, and a set of windows you can show, hide, dock, or float around the canvas.

info

This page describes the current editor, which is the default for all users. A legacy editor is still available — see Switching to the legacy editor below. The legacy layout is documented in the Studio UI Guide.

Menu bar

The menu bar spans the top of the editor. On the left it shows the TangleML logo (a link back to your dashboard), the current pipeline name, and the menus. On the right it shows the run controls.

Hover the pipeline name to reveal a pencil button that opens the Rename Pipeline dialog.

File

Item	Description
Open	Open another pipeline without leaving the editor
Save	Save the current pipeline
Save as	Save a copy under a new name
Rename	Rename the current pipeline
New pipeline	Start a fresh, empty pipeline
Import	Load a pipeline from a file or by pasting YAML
Export	Download the pipeline as a YAML file
Move to folder	Move the pipeline into a pipeline folder
Delete pipeline	Delete the pipeline, after a confirmation prompt

View

• Auto-layout — re-arrange the graph using one of four algorithms: Sugiyama, Sugiyama Centered, Compact, or Spread. Sugiyama is bound to a keyboard shortcut, shown next to the item.
• View YAML — open the full pipeline definition in a fullscreen, read-only viewer.
• Focus mode — a checkbox that hides surrounding chrome so you can concentrate on the canvas. Also available via a keyboard shortcut.

Runs

• Submit Run — submit the pipeline as-is.
• Submit with Arguments — open a dialog to review and set argument values before submitting.

Components

• Explore library — open the component browser.
• Add component — add a new component to your user library.

Windows

The Windows menu lists every editor window with a checkbox, so you can show or hide each one individually. A Views submenu applies a whole layout at once:

Preset	Effect
Default	The standard working layout
All	Every window visible, including diagnostics panels
Minimal	Hide all windows, leaving only the canvas

Node

The Node menu appears only while something is selected on the canvas, and its items depend on what that something is:

Item	Available for
Duplicate Task	Tasks
Arrange	Any node — Move Forward, Move Backward, Bring to Front, Send to Back
Unpack Subgraph	Subgraph nodes — replaces the node with its constituent tasks
Delete Task	Tasks

Arrange controls stacking order, which matters when nodes overlap — most often when a sticky note sits behind or in front of the tasks it annotates.

Run controls

At the right of the menu bar:

• Submit Run — a quick-run button. It is accompanied by a Submit run with arguments option. When the pipeline has validation problems, the button reports them — either Pipeline has validation errors (which blocks submission) or Pipeline has validation warnings.
• Auto-save indicator — shows Auto-save enabled normally and Saving… while a save is in flight, so you can tell your work has been persisted without saving manually.

Keyboard shortcuts

Shortcuts use Cmd on macOS and Alt elsewhere; either modifier works, so the tables below write it as Cmd/Alt.

Shortcut	Action
Cmd/Alt + /	Toggle focus mode
Cmd/Alt + Shift + L	Auto-layout the graph
Cmd/Alt + C	Copy the current selection
Cmd/Alt + V	Paste at the pointer
Cmd/Alt + Z	Undo
Cmd/Alt + Shift + Z	Redo
Cmd/Alt + Y	Redo (alternative)
Ctrl + O	Open a pipeline
Ctrl + Shift + M	Maximize the arguments editor in the Properties window
Delete or Backspace	Delete the current selection
Esc	Dismiss the topmost overlay, or clear the canvas selection

Esc works in layers: if a dialog, popover, or menu is open it closes that first, and only clears your canvas selection once nothing is stacked on top.

Windows

Editor panels are windows rather than fixed sidebars. Each window can be docked to the left or right edge, floated over the canvas, minimized, maximized, or hidden. Your layout is saved in the browser and restored on your next visit.

Window controls

Every window header carries a set of buttons:

Control	Description
Minimize / Expand	Collapse a docked window to its title bar, or expand it again
Maximize / Restore	Expand a window to fill the available area, or return it to its previous size
Hide	Remove the window from view; bring it back from the Windows menu

Drag a window by its header to move it. Dragging toward the left or right edge docks it there; dragging away from an edge floats it over the canvas. Docked windows on the same side stack vertically.

Available windows

Window	Contents
Component Search	Search for components across every available source and drop them onto the canvas. See Component Search
Runs & Submissions	Submit the current pipeline. If you are not signed in, this window offers a sign-in button instead
Recent Runs	Recent runs for this pipeline, for quick navigation
Pipeline Details	Pipeline-level information: name, timestamps, digest, validation status, tags, and inputs and outputs. When you are inside a subgraph the window is titled Subgraph Details and describes that subgraph instead
Properties	Context panel for the current selection. See Task properties
Pipeline Structure	A tree view of the pipeline, including nested subgraphs and per-entity issue badges. See Pipeline Structure
Fix Issues	Guided fixes for validation problems, one issue at a time. See Validation and fixing issues
History	The undo/redo history for this editing session. See History
Minimap	An overview map of the whole graph plus zoom, fit-view, and lock controls
Tip of the Day	A rotating usage tip
Debug Panel	Diagnostic information about editor state, useful when reporting a problem. Hidden in the Default view; show it from the Windows menu or the All view

Canvas

The canvas is your workspace for building the pipeline graph. Tasks appear as nodes with inputs on the left and outputs on the right; drag from an output handle to an input handle to connect them.

Alongside task nodes, the editor supports pipeline input and output nodes for pipeline-level parameters and results, and subgraph nodes for nested pipelines.

Use the Minimap window for navigation and the View → Auto-layout menu to tidy up the graph.

Adding to the canvas

There are several ways to get something onto the canvas:

• Drag a component from the Component Search window.
• Drop a component YAML file straight onto the canvas. The component is added to the canvas and to your user component library at the same time.
• Paste a copied selection with Cmd/Alt + V; it lands at the pointer.
• Drag Sticky Note, Input, or Output out of the Component Search window's browse view — see Component Search.

Replacing a task's component

Drop a component onto an existing task node to swap out its component reference. Before anything changes, a Replace Task? dialog shows which inputs and outputs the new component adds, keeps, or removes. If any input or output disappears, the dialog warns that their connections will be lost. Continue applies the swap; Cancel leaves the task untouched.

Working with a multiple selection

Hold Shift and drag to draw a selection box, or hold Cmd/Ctrl and click nodes to add them one at a time. A small toolbar appears next to the selection:

Button	Effect
Duplicate	Copy the selected nodes in place
Copy	Copy the selection to the clipboard
Create Subgraph	Collapse the selection into a subgraph — shown for two or more tasks
Delete	Remove the selected nodes

Create Subgraph opens a small form asking for a subgraph name before it collapses anything.

With a multiple selection active, the Properties window switches to a batch view showing how many nodes are selected, a list of them, and:

• Task color — recolor every selected task at once.
• Common Arguments — arguments that all the selected tasks share. Editing one applies to the whole selection; where the tasks currently disagree, the field reads mixed and hovering it explains that values differ across tasks.
• Auto Layout Selection — re-arrange just the selected nodes, leaving the rest of the graph alone.
• Create Subgraph — the same action as the floating toolbar.

Sticky notes

Sticky notes are annotation nodes that carry no execution meaning — use them to label regions of a graph or leave a note for a collaborator. Drag Sticky Note from the Canvas Tools folder in the Component Search window's browse view.

Select a note to edit it in the Properties window:

• Title and content, each with its own font size.
• Background colour and border colour.
• A lock toggle that freezes the note's text and colours so you cannot edit it by accident.
• Read-only Layout (size, position, z-index) and Metadata (id, created timestamp, author) values, each copyable.

Notes are resizable by dragging their edges, and their stacking order is controlled from Node → Arrange.

Guidelines

Guidelines are horizontal or vertical lines you can route edges along, to keep a dense graph readable. They are created from the keyboard:

• Hold Shift + Q and click the canvas for a horizontal guideline.
• Hold Shift + W and click the canvas for a vertical guideline.

The new guideline is selected immediately, and the Properties window shows its colour, its orientation, and the list of edges assigned to it. While a guideline is selected the canvas dims unassigned edges and highlights assigned ones in the guideline's colour: click an edge to assign it, click it again to unassign. Each guideline gets a distinct colour automatically.

Drag a guideline to move it; the edges assigned to it follow. Delete it from the Properties window, and the edges revert to their normal routing.

Task properties

Select a task to populate the Properties window. It has three tabs.

Arguments

Every input the component declares, with its type and its current value. Each row has a lightning-bolt button that opens a menu of ways to fill the argument:

Item	Effect
Reset to Default	Restore the component's declared default
Unset Argument	Clear the value entirely
Dynamic Data	Resolve the value at run time rather than hard-coding it — see below
Connect to…	Wire the input to a matching upstream output, grouped by source node and labelled with each port's type
Create Input	Create a pipeline-level input for this argument and connect it in one step

Dynamic Data covers values that should not be baked into the pipeline definition:

• Select Secret… opens a picker over the secrets available to you, so the pipeline stores a reference rather than the secret itself. See Secrets.
• The remaining entries are values the platform supplies at run time.

For long or structured values, the arguments editor opens a code editor with syntax highlighting. It picks a language from the argument's type and lets you override it — JSON, YAML, Python, JavaScript, TypeScript, HTML, CSS, SQL, Markdown, Shell, or plain text. Ctrl + Shift + M maximizes the editor while you work in it.

Alongside the tabs, arguments show a validation indicator when a required value is missing or a connection no longer resolves.

Config

Execution settings for the selected task:

• Task color — a colour for the node on the canvas, which also drives the batch recolour in a multiple selection.
• Disable cache — opt this task out of caching. Not shown for subgraph tasks. See Caching.
• Collapse node — draw the node in a condensed form on the canvas.
• Compute resources — cloud provider and per-provider resource settings. Choosing a provider swaps in that provider's fields; switching provider clears the fields belonging to the old one.

Annotations

Key/value metadata attached to the task, with your own annotations kept separate from the ones the platform maintains.

Beneath the tabs, an Actions section collects the operations available for the selected task.

Validation and fixing issues

The editor validates the pipeline continuously. The Pipeline Details window summarises the result as, for example, 2 errors, 1 warning. Click the summary to expand the full list, or click Fix to open the Fix Issues window.

Fix Issues walks the pipeline's issues one at a time. A header shows your position (3 of 7) with Previous issue and Next issue buttons; moving between issues also navigates the canvas to the entity concerned, including into a subgraph when the issue is nested. When everything validates, the window reads No issues to fix.

Each issue shows its severity and message, followed by whatever fix applies to it:

Problem	Offered fix
A pipeline input has no value	Enter the value inline
A required task input is unset	Supply the input
A task, component, input, or output has an empty name	Rename it inline
Two inputs or two outputs share a name	Rename one of them
A task has no component reference	Delete Task
An input or output is unconnected	Delete Input / Delete Output
A connection points at something that no longer exists	Clear the broken reference
A connection is left dangling at one end	Delete Binding
The component could not be loaded from its source	Guidance to check the component URL or your connection, then reload
A required launcher annotation is missing	Guidance to fill it in on the task's Config tab
The graph contains a circular dependency	Guidance — cycles have to be broken by hand
The pipeline has no tasks	Guidance to add tasks from the component library

Issues that cannot be fixed automatically say so rather than offering a button that would not help.

Pipeline Structure

The Pipeline Structure window is a tree of the pipeline: the root pipeline, its tasks, and its subgraphs nested inside one another. The branch you are currently editing is highlighted, and the tree expands to follow you as you navigate into subgraphs.

Entities with validation problems carry a badge — red with a total count when there are errors, amber with a count when there are only warnings — so you can see which branch of a large pipeline needs attention without expanding all of it. Expanding an entity lists its issues; clicking one selects it and navigates the canvas to it, the same way the Fix Issues window does.

History

The History window lists every change in this editing session, most recent last, with the initial state marked at the top. Its toolbar has undo and redo buttons, a counter reading undone / total, and a Clear history button that discards the recorded history.

Undo history is scoped to the pipeline you are editing. The most recent changes are saved in the browser and restored when you reopen the pipeline, so a reload does not cost you your last few undos.

Subgraphs

A group of tasks can be collapsed into a subgraph node, which keeps large pipelines readable. Select two or more tasks and use Create Subgraph from the floating selection toolbar or the Properties window.

Double-click a subgraph node to navigate into it; the Pipeline Details window becomes Subgraph Details while you are inside, and the Pipeline Structure window shows where you are in the nesting.

To flatten a subgraph back into its individual tasks, select it and choose Node → Unpack Subgraph.

Opening a pipeline

Landing in the editor without a pipeline shows an Open Pipeline panel with your pipeline folders, so you can pick one without going back to the dashboard. The same picker is available at any time from File → Open or with Ctrl + O.

Switching to the legacy editor

A version toggle in the editor lets you move between the current and legacy editors for the same pipeline. Hover it to see Switch to legacy editor; from the legacy editor the same control reads Switch to new editor. Both editors read and write the same pipeline, so you can switch back and forth without losing work.

The first time you land in the current editor, a short welcome message points out the toggle. Dismiss it and it will not reappear.