latest--partial-step.schema

Breaks out of the current loop (e.g., WHILE or FOR) based on the given condition.

Optional user-friendly label shown in the Flow Monitor for this step. Supports variable resolution (e.g., "Translating to {{ flowContext.targetLang }}..."). Falls back to the step name if not provided.

The condition to evaluate. Loop will be broken when condition evaluates to true. If not provided, breaks unconditionally.

Logs debug information about the current flow context and step configuration. Optionally triggers a debugger breakpoint.

Optional user-friendly label shown in the Flow Monitor for this step. Supports variable resolution (e.g., "Translating to {{ flowContext.targetLang }}..."). Falls back to the step name if not provided.

Optional identifier for the debug step, included in the log message.

If true, skips triggering the debugger breakpoint.

Emits a custom internal event that can trigger other actions subscribed via action events. (Note: Aliases DISPATCH_EVENT are deprecated, use 'EMIT_EVENT' instead)

Event name to emit. Supports variable resolution.

Optional user-friendly label shown in the Flow Monitor for this step. Supports variable resolution (e.g., "Translating to {{ flowContext.targetLang }}..."). Falls back to the step name if not provided.

Optional event payload object. Supports variable resolution.

Executes a set of steps for a fixed number of iterations (numeric) or iterates over an array (items).

The name of the variable to set with the current iteration value.

Optional user-friendly label shown in the Flow Monitor for this step. Supports variable resolution (e.g., "Translating to {{ flowContext.targetLang }}..."). Falls back to the step name if not provided.

The starting value of the loop variable (numeric).

The ending value of the loop variable (numeric, inclusive).

Step increment for numeric loops (can be negative). If omitted, inferred from start/end (1 or -1).

An array to iterate over (e.g., an expression producing an array). If provided, numeric start/end are ignored.

Steps to execute in each iteration of the loop.

Execute a conditional branch based on a condition

Optional user-friendly label shown in the Flow Monitor for this step. Supports variable resolution (e.g., "Translating to {{ flowContext.targetLang }}..."). Falls back to the step name if not provided.

The condition to evaluate. Can be a template expression with {{ }} syntax.

Legacy alternative to condition. The condition to evaluate using JavaScript syntax.

Steps to execute if the condition is true

Steps to execute if the condition is false

Ends the execution of a flow based on the given condition, with optional notification. (Note: Aliases END_IF are deprecated, use 'END' instead)

Optional user-friendly label shown in the Flow Monitor for this step. Supports variable resolution (e.g., "Translating to {{ flowContext.targetLang }}..."). Falls back to the step name if not provided.

The condition to evaluate. Flow will be terminated gracefully when condition evaluates to true.

Optional notification to show when ending the flow.

Runs multiple step lists in parallel and merges their flow contexts upon completion.

An array of step arrays, where each sub-array represents a branch of steps to execute in parallel.

Optional user-friendly label shown in the Flow Monitor for this step. Supports variable resolution (e.g., "Translating to {{ flowContext.targetLang }}..."). Falls back to the step name if not provided.

Routes flow execution to platform-specific branches (swing, prime, standalone) with optional default fallback.

Optional user-friendly label shown in the Flow Monitor for this step. Supports variable resolution (e.g., "Translating to {{ flowContext.targetLang }}..."). Falls back to the step name if not provided.

Steps executed when running in Swing.

Steps executed when running in Prime.

Steps executed when running in Standalone mode.

Fallback steps executed when no platform-specific branch matches.

Sets variables in the flow context. Can execute steps, evaluate expressions, or set key-value pairs.

Optional user-friendly label shown in the Flow Monitor for this step. Supports variable resolution (e.g., "Translating to {{ flowContext.targetLang }}..."). Falls back to the step name if not provided.

Steps to execute before setting the variable. The result is stored in the flow context.

JavaScript expression to evaluate and set as the variable value.

Text to resolve (with variables) and set as the variable value.

Raw text to set as the variable value without resolution.

Name of the variable to set in the flow context.

Convert the value to a specific type (e.g., "number").

Output mappings from step outputs to flow context

Evaluates a switch value and executes the steps for the matching case. Supports both array and object syntax for cases.

The cases to match against the switch value. Each case key should match a possible switch value, with the value being an array of steps to execute. Use "default" as a fallback case.

Optional user-friendly label shown in the Flow Monitor for this step. Supports variable resolution (e.g., "Translating to {{ flowContext.targetLang }}..."). Falls back to the step name if not provided.

The condition to evaluate for the switch value. Can be a template expression with {{ }} syntax.

Legacy alternative to condition. The switch value to evaluate using JavaScript syntax.

Calls a reusable flow by legacy template name (name) or scoped reference (ref).

Optional user-friendly label shown in the Flow Monitor for this step. Supports variable resolution (e.g., "Translating to {{ flowContext.targetLang }}..."). Falls back to the step name if not provided.

The name of the template to call from the TEMPLATES: section.

Reference path to a reusable flow (for example global.flows.myFlow or imports.common.flows.myFlow).

Executes a set of steps in a try block, and if an error occurs, executes catch steps.

Steps to execute in the try block.

Optional user-friendly label shown in the Flow Monitor for this step. Supports variable resolution (e.g., "Translating to {{ flowContext.targetLang }}..."). Falls back to the step name if not provided.

Steps to execute if an error occurs in the try block.

Repeatedly executes a set of steps while a condition evaluates to true.

Optional user-friendly label shown in the Flow Monitor for this step. Supports variable resolution (e.g., "Translating to {{ flowContext.targetLang }}..."). Falls back to the step name if not provided.

The condition to evaluate for the loop. Can be a template expression with {{ }} syntax.

Legacy alternative to condition. The test expression to evaluate using JavaScript syntax.

Steps to execute in each iteration of the loop.

Removes all occurrences of the specified tags from the document or report while preserving their text content. This is useful for cleaning up temporary markup or removing specific tag types.

Optional user-friendly label shown in the Flow Monitor for this step. Supports variable resolution (e.g., "Translating to {{ flowContext.targetLang }}..."). Falls back to the step name if not provided.

Array of tag names to clear (e.g., ["mark", "highlight"]). Can be a hardcoded array, a flowContext variable, or read from the default list input.

Input mappings from flow context to step inputs

Closes the current document without saving.

Optional user-friendly label shown in the Flow Monitor for this step. Supports variable resolution (e.g., "Translating to {{ flowContext.targetLang }}..."). Falls back to the step name if not provided.

Saves the current document. Optionally closes the document after saving.

Optional user-friendly label shown in the Flow Monitor for this step. Supports variable resolution (e.g., "Translating to {{ flowContext.targetLang }}..."). Falls back to the step name if not provided.

Whether to close the document after saving. Supports variable resolution.

Retrieves the full pure text content from the current document and sets it as output.

Optional user-friendly label shown in the Flow Monitor for this step. Supports variable resolution (e.g., "Translating to {{ flowContext.targetLang }}..."). Falls back to the step name if not provided.

Output mappings from step outputs to flow context

Retrieves zones and their linked documents from a DWP or Report.

Optional user-friendly label shown in the Flow Monitor for this step. Supports variable resolution (e.g., "Translating to {{ flowContext.targetLang }}..."). Falls back to the step name if not provided.

Whether to include linked documents in each zone. Defaults to true.

Output mappings from step outputs to flow context

Retrieves text content from the document or a specific location and sets it as output. (Note: Aliases CTX_GET_TEXT_CONTENT are deprecated, use 'GET_TEXT_CONTENT' instead)

Optional user-friendly label shown in the Flow Monitor for this step. Supports variable resolution (e.g., "Translating to {{ flowContext.targetLang }}..."). Falls back to the step name if not provided.

Location to read text from. Options: CURSOR, CURSOR_PARAGRAPH, XPATH, REPORT.

XPath expression for reading text at a specific location.

Output mappings from step outputs to flow context

Retrieves XML content from the document or a specific location and sets it as output. (Note: Aliases CTX_GET_XML_CONTENT are deprecated, use 'GET_XML_CONTENT' instead)

Optional user-friendly label shown in the Flow Monitor for this step. Supports variable resolution (e.g., "Translating to {{ flowContext.targetLang }}..."). Falls back to the step name if not provided.

Location to read XML from. Options: CURSOR, CURSOR_PARAGRAPH, XPATH, REPORT.

XPath expression for reading XML at a specific location.

Output mappings from step outputs to flow context

Retrieves information about the currently selected content object and sets it as output. (Note: Aliases CTX_SELECTED_OBJECT are deprecated, use 'SELECTED_OBJECT' instead)

Optional user-friendly label shown in the Flow Monitor for this step. Supports variable resolution (e.g., "Translating to {{ flowContext.targetLang }}..."). Falls back to the step name if not provided.

Output mappings from step outputs to flow context

Insert object references or templates into the current document using a DTX template.

Prime: Supports paths, UUIDs, and LOIDs (auto-formatted). Options insertionType, full, and templateType are ignored (warnings logged).

Optional user-friendly label shown in the Flow Monitor for this step. Supports variable resolution (e.g., "Translating to {{ flowContext.targetLang }}..."). Falls back to the step name if not provided.

Optional input variable. String values are treated as path, arrays as referenceListIds.

Single object path to insert (convenience option).

List of object identifiers (paths, loids, or uuids).

List of full object entries to insert when full is true.

If true, uses referenceListObjects as full object payload.

When using IDs, indicates that identifiers are paths.

Insertion type for Swing (insert, embed, link). Prime ignores this option.

Template name to use for insertion.

Template type (Swing only).

Write content even if the document is considered read-only.

Input mappings from flow context to step inputs

Set a field value in the object panel using a selector or XPath. Supports tagsinput and plain fields. (Note: Aliases CTX_SET_METADATA are deprecated, use 'SET_METADATA' instead)

CSS selector of the input element to update.

Optional user-friendly label shown in the Flow Monitor for this step. Supports variable resolution (e.g., "Translating to {{ flowContext.targetLang }}..."). Falls back to the step name if not provided.

Input mappings from flow context to step inputs

Shows a response to the user in a modal. Can run inline steps before showing and supports HTML and Markdown modes. (Note: Aliases CTX_SHOW_RESPONSE are deprecated, use 'SHOW_RESPONSE' instead)

Optional user-friendly label shown in the Flow Monitor for this step. Supports variable resolution (e.g., "Translating to {{ flowContext.targetLang }}..."). Falls back to the step name if not provided.

Steps to execute while showing an in-place loading UI before the response is displayed.

Render mode for the response. Use "html" to render HTML, "markdown" to render Markdown, or "text" for plain text.

Input mappings from flow context to step inputs

Shows a notification to the user with optional title and resolved message. (Note: Aliases CTX_SHOW_NOTIFICATION are deprecated, use 'SHOW_NOTIFICATION' instead)

The message to show. Supports variable resolution.

Optional user-friendly label shown in the Flow Monitor for this step. Supports variable resolution (e.g., "Translating to {{ flowContext.targetLang }}..."). Falls back to the step name if not provided.

Type of notification (info, warning, error, message, task, success).

Optional title for the notification.

Insert text at the given location (cursor, xpath or report). If the step reads text from the flowContext, you can use a preceding SET step to provide the content. (Note: Aliases CTX_INSERT_TEXT are deprecated, use 'INSERT_TEXT' instead)

Optional user-friendly label shown in the Flow Monitor for this step. Supports variable resolution (e.g., "Translating to {{ flowContext.targetLang }}..."). Falls back to the step name if not provided.

Optional path or variable name to read the content from the flowContext. If omitted, the step uses the content from the default text input.

Where to insert the text. Use CURSOR (default), XPATH, or REPORT.

XPath expression used when at is XPATH.

Write content even if the document is considered read-only.

Input mappings from flow context to step inputs

Insert XML content at the given location. The content can be provided via the flowContext (using in) or via the default text input. (Note: Aliases CTX_INSERT_XML are deprecated, use 'INSERT_XML' instead)

Optional user-friendly label shown in the Flow Monitor for this step. Supports variable resolution (e.g., "Translating to {{ flowContext.targetLang }}..."). Falls back to the step name if not provided.

Optional path or variable name to read the XML content from the flowContext. If omitted, the step uses the default text input.

Where to insert the XML. Use CURSOR (default) or XPATH.

XPath expression used when at is XPATH.

Position mode for insertion when using cursor (e.g. insertInline, insertBefore, insertAfter).

Write content even if the document is considered read-only.

Replace existing text at the given location (cursor paragraph, cursor or xpath) with the provided content. Content can come from a flow variable (using in) or the default text input. (Note: Aliases CTX_REPLACE_TEXT are deprecated, use 'REPLACE_TEXT' instead)

Optional user-friendly label shown in the Flow Monitor for this step. Supports variable resolution (e.g., "Translating to {{ flowContext.targetLang }}..."). Falls back to the step name if not provided.

Optional text or variable name to read the content from the flowContext. If omitted, the step uses the default text input.

Where to replace the text. Options: XPATH, CURSOR_PARAGRAPH, CURSOR.

XPath expression used when at is XPATH.

Write content even if the document is considered read-only.

Replace XML content at the given location. Content can be sourced from a flow variable (in) or the default text input. (Note: Aliases CTX_REPLACE_XML are deprecated, use 'REPLACE_XML' instead)

Optional user-friendly label shown in the Flow Monitor for this step. Supports variable resolution (e.g., "Translating to {{ flowContext.targetLang }}..."). Falls back to the step name if not provided.

Optional text or variable name to read the XML content from the flowContext.

Where to replace the XML. Options: XPATH, CURSOR_PARAGRAPH, CURSOR.

XPath expression used when at is XPATH.

Write content even if the document is considered read-only.

Insert a list of items into the document. The items can be provided as a list input or be generated from a text input (converted to a list using ProcessToList). (Note: Aliases CTX_INSERT_LIST are deprecated, use 'INSERT_LIST' instead)

Optional user-friendly label shown in the Flow Monitor for this step. Supports variable resolution (e.g., "Translating to {{ flowContext.targetLang }}..."). Falls back to the step name if not provided.

Optional text or variable name to read the content (text) from the flowContext. If omitted, the step uses the default text input.

Container tag to use for the list (default 'ul').

List item tag to use for entries (default 'li').

When true, inserts items one-by-one instead of a full container.

XPath expression used when at is XPATH.

Where to insert the list. Use CURSOR (default) or XPATH.

Position mode for insertion when using cursor (e.g. insertInline, insertBefore, insertAfter).

Write content even if the document is considered read-only.

Input mappings from flow context to step inputs

Adds tags to pure text content in the document. Tags can highlight or annotate specific text ranges. In Swing, tags can be virtual (temporary) or persistent. In Prime, tags are always persistent.

Optional user-friendly label shown in the Flow Monitor for this step. Supports variable resolution (e.g., "Translating to {{ flowContext.targetLang }}..."). Falls back to the step name if not provided.

Array of tags to add. Each tag must have: tag (tag name), start (position), size (length), and optional attributes (key-value pairs). Can be a hardcoded array, a flowContext variable, or read from the default list input.

Whether tags should be virtual (temporary, not saved). Default is false (persistent tags). Only applies to Swing; Prime always persists tags.

Input mappings from flow context to step inputs

Creates a new section in the document using the specified template path, relative to the currently elected item

The path to the template to be inserted (supports variables)

Optional user-friendly label shown in the Flow Monitor for this step. Supports variable resolution (e.g., "Translating to {{ flowContext.targetLang }}..."). Falls back to the step name if not provided.

Whether the section should be inserted after the current section (true) or before (false). Defaults to true

Display a selection of choices in the Flow Monitor. Pauses workflow until user selects.

The prompt message asking user to make a selection.

Array of choices. Each choice has: id (returned value), label (display text), optional description.

Optional user-friendly label shown in the Flow Monitor for this step. Supports variable resolution (e.g., "Translating to {{ flowContext.targetLang }}..."). Falls back to the step name if not provided.

ID (or array of IDs in multi mode) of default/pre-selected choice(s).

Selection mode: single (default) or multi.

Minimum selections required in multi mode.

Maximum selections allowed in multi mode.

Optional custom text input below options.

If true, shows a Cancel option that returns empty string.

Output mappings from step outputs to flow context

Display a confirmation dialog in the Flow Monitor. Pauses workflow until user responds.

The confirmation message to display. Supports variable resolution with {{variable}}.

Optional user-friendly label shown in the Flow Monitor for this step. Supports variable resolution (e.g., "Translating to {{ flowContext.targetLang }}..."). Falls back to the step name if not provided.

Optional rich content to display (e.g., AI-generated text for review). Supports markdown.

Label for the confirm button (default: "Confirm"). Used in Yes/No mode.

Label for the cancel button (default: "Cancel"). Used in Yes/No mode.

Visual style of the confirm button.

Optional timeout in seconds. Auto-cancels if user does not respond.

Output mappings from step outputs to flow context

Display an inline text input field in the Flow Monitor.

The prompt message asking for input.

Optional user-friendly label shown in the Flow Monitor for this step. Supports variable resolution (e.g., "Translating to {{ flowContext.targetLang }}..."). Falls back to the step name if not provided.

Placeholder text shown in the input field.

Default value pre-filled in the input field.

If true, empty input is not allowed.

Maximum number of characters allowed.

Error message shown when the field is submitted empty and required is true.

Output mappings from step outputs to flow context

Display JSON structured data in the Flow Monitor.

Object or array data to render as JSON.

Optional user-friendly label shown in the Flow Monitor for this step. Supports variable resolution (e.g., "Translating to {{ flowContext.targetLang }}..."). Falls back to the step name if not provided.

Optional title displayed above JSON.

Maximum depth for JSON rendering.

Display key-value structured data in the Flow Monitor.

Object to render as key-value pairs.

Optional user-friendly label shown in the Flow Monitor for this step. Supports variable resolution (e.g., "Translating to {{ flowContext.targetLang }}..."). Falls back to the step name if not provided.

Optional title displayed above key-value pairs.

Display metrics in the Flow Monitor.

Metrics as object or array of label/value pairs.

Optional user-friendly label shown in the Flow Monitor for this step. Supports variable resolution (e.g., "Translating to {{ flowContext.targetLang }}..."). Falls back to the step name if not provided.

Optional title displayed above metrics.

Display status messages in the Flow Monitor during workflow execution.

Optional user-friendly label shown in the Flow Monitor for this step. Supports variable resolution (e.g., "Translating to {{ flowContext.targetLang }}..."). Falls back to the step name if not provided.

The status message to display. Supports variable resolution with {{variable}}.

Type of status message.

If true, clears all previous messages before showing this one.

Display tabular structured data in the Flow Monitor.

Array of objects to render as a table.

Optional user-friendly label shown in the Flow Monitor for this step. Supports variable resolution (e.g., "Translating to {{ flowContext.targetLang }}..."). Falls back to the step name if not provided.

Optional title displayed above the table.

Maximum number of rows rendered before truncation.

Triggers the digital preview functionality

Optional user-friendly label shown in the Flow Monitor for this step. Supports variable resolution (e.g., "Translating to {{ flowContext.targetLang }}..."). Falls back to the step name if not provided.

Executes an array of commands in the document preview panel (e.g. CREATE_PDF, CREATE_TOC)

Array of strings representing commands to execute (e.g. ["CREATE_TOC", "CREATE_PDF"])

Optional user-friendly label shown in the Flow Monitor for this step. Supports variable resolution (e.g., "Translating to {{ flowContext.targetLang }}..."). Falls back to the step name if not provided.

Triggers the print design preview functionality

Optional user-friendly label shown in the Flow Monitor for this step. Supports variable resolution (e.g., "Translating to {{ flowContext.targetLang }}..."). Falls back to the step name if not provided.

Sanitizes text input. Can strip markdown code blocks, normalize Unicode characters (quotes, dashes, spaces), and validate/repair XML content before forwarding the result to the flow context.

Optional user-friendly label shown in the Flow Monitor for this step. Supports variable resolution (e.g., "Translating to {{ flowContext.targetLang }}..."). Falls back to the step name if not provided.

If true, removes markdown code blocks (...) from the input before further processing.

If true, attempts to validate and repair XML fragments found in the input.

If true, normalizes Unicode characters to their standard equivalents (e.g., curly quotes to straight quotes, various dashes to hyphens, non-breaking spaces to regular spaces). This ensures consistency between AI responses and input text.

Input mappings from flow context to step inputs

Output mappings from step outputs to flow context

Executes inline JavaScript (script) or a JS-style template (template). Use with caution — only in trusted flows.

Optional user-friendly label shown in the Flow Monitor for this step. Supports variable resolution (e.g., "Translating to {{ flowContext.targetLang }}..."). Falls back to the step name if not provided.

A JavaScript expression or block to be executed in the step context. If mode: "async" the script is treated as async.

A template string that will be resolved using the step context and then parsed as JSON. Use this when you want to build structured output.

Reference path to a reusable script (for example local.scripts.normalizeName or imports.common.scripts.normalizeName).

Execution mode for scripts. Use "async" to run the script using the asynchronous evaluator.

When specified, the script is treated as an expression (no return statement needed) and the result is assigned to the specified property in flowContext. Cannot be used with template mode.

Convert a textual AI response into a list of strings. Detects JSON arrays, ordered/unordered lists, comma-separated values, or falls back to a single-item array. (Note: Aliases PROCESS_TO_LIST are deprecated, use 'TO_LIST' instead)

Optional user-friendly label shown in the Flow Monitor for this step. Supports variable resolution (e.g., "Translating to {{ flowContext.targetLang }}..."). Falls back to the step name if not provided.

Input mappings from flow context to step inputs

Output mappings from step outputs to flow context

Parses JSON content from the text input and stores the resulting object in the flow context. (Note: Aliases PROCESS_PARSE_JSON are deprecated, use 'PARSE_JSON' instead)

Optional user-friendly label shown in the Flow Monitor for this step. Supports variable resolution (e.g., "Translating to {{ flowContext.targetLang }}..."). Falls back to the step name if not provided.

Optional path or variable name to read the JSON string from the flowContext. If omitted, the default text input is used.

Input mappings from flow context to step inputs

Output mappings from step outputs to flow context

Converts markdown text to HTML and sets it as text output.

Optional user-friendly label shown in the Flow Monitor for this step. Supports variable resolution (e.g., "Translating to {{ flowContext.targetLang }}..."). Falls back to the step name if not provided.

Text input or flow variable to convert. If omitted, the default text input is used.

Converts a base64-encoded string to a Blob and stores it in the flow context as a blob output.

Optional user-friendly label shown in the Flow Monitor for this step. Supports variable resolution (e.g., "Translating to {{ flowContext.targetLang }}..."). Falls back to the step name if not provided.

Optional path or variable name to read the base64 string from the flowContext. If omitted, the default text input is used.

Optional content type for the resulting Blob (e.g. "image/png").

Output mappings from step outputs to flow context

Reads content from the user clipboard. Prefers HTML content when available, falls back to plain text.

Optional user-friendly label shown in the Flow Monitor for this step. Supports variable resolution (e.g., "Translating to {{ flowContext.targetLang }}..."). Falls back to the step name if not provided.

Output mappings from step outputs to flow context

Writes text to the user clipboard. Reads the content from the configured input (cfg.text) or from the default text input in the flow context.

Optional user-friendly label shown in the Flow Monitor for this step. Supports variable resolution (e.g., "Translating to {{ flowContext.targetLang }}..."). Falls back to the step name if not provided.

Text to write to clipboard. If omitted, uses the default text input from flow context.

Downloads a file constructed from text or an existing Blob. The filename is required. (Note: Aliases DOWNLOAD_TEXT are deprecated, use 'DOWNLOAD' instead)

Optional user-friendly label shown in the Flow Monitor for this step. Supports variable resolution (e.g., "Translating to {{ flowContext.targetLang }}..."). Falls back to the step name if not provided.

Name of the file to download (required).

The content-type to set for the download of the file.

Input mappings from flow context to step inputs

Opens a file upload dialog (modal) and stores selected file(s) into the flow context. Supports output mapping via the outputs config or default file output. (Note: Aliases USER_FILE_UPLOAD are deprecated, use 'FILE_UPLOAD' instead)

Optional user-friendly label shown in the Flow Monitor for this step. Supports variable resolution (e.g., "Translating to {{ flowContext.targetLang }}..."). Falls back to the step name if not provided.

Prompt text shown in the upload dialog.

Comma-separated list of accepted MIME types for upload (e.g. "image/png,application/pdf").

Array of accepted MIME types for the upload (e.g. ["audio/mp3","audio/wav"]).

Maximum file size allowed in bytes.

Output mappings from step outputs to flow context

Shows a simple prompt to the user and stores the entered text into the default text output. (Note: Aliases USER_PROMPT are deprecated, use 'PROMPT' instead)

Optional user-friendly label shown in the Flow Monitor for this step. Supports variable resolution (e.g., "Translating to {{ flowContext.targetLang }}..."). Falls back to the step name if not provided.

Text displayed in the prompt. Can contain template placeholders resolved against the flowContext.

Optional placeholder shown inside the input field.

Output mappings from step outputs to flow context

Displays a configurable form to the user (via FormBuilder). The form definition is provided in cfg.form. When submitted, the returned values are merged into the flowContext.

Form field definitions. Each key is the field name, and the value is a FormComponent object defining the field type and properties. See the comprehensive form documentation for all available field types and properties.

Optional user-friendly label shown in the Flow Monitor for this step. Supports variable resolution (e.g., "Translating to {{ flowContext.targetLang }}..."). Falls back to the step name if not provided.

Title of the form modal.

Optional buttons configuration for the form modal. Array of button objects defining submit/cancel buttons.

Optional modal-level configuration (width, height, fullScreen, typography). See FormModalConfig for all options.

Optional steps to execute before showing the form while an in-place loading UI is shown.

Plays audio in a lightweight floating player. Audio source can be provided as a data URL, http(s) URL, or raw base64 via cfg.in or via the default text input. The step mounts a closed Shadow DOM audio player and resolves when the player is closed by the user.

Optional user-friendly label shown in the Flow Monitor for this step. Supports variable resolution (e.g., "Translating to {{ flowContext.targetLang }}..."). Falls back to the step name if not provided.

Optional input path or variable name containing the audio source (data URL, http(s) URL or raw base64). If omitted, the default text input is used.

Attempt autoplay on load. Note: browsers may block autoplay; in that case the player will show a hint and wait for user interaction.

Show progress / seek bar in the player.

Show volume controls in the player.

Show current time and duration labels.

Theme of the player: 'dark' | 'light' or any valid CSS color string.

Player anchoring position: 'top-left' | 'top-right' | 'bottom-left' | 'bottom-right' | 'top-center' | 'bottom-center' | 'center'.

If true, loop playback.

Optional title shown in the player title bar.

Optional MIME type hint used when the provided input is raw base64 without a data URL header (e.g. "audio/mpeg").

Input mappings from flow context to step inputs

Displays a selection modal allowing the user to pick one item from a list. The list may be provided directly or produced by a pre-processing step (inlineSteps). The selected value is written to the default text output.

Optional user-friendly label shown in the Flow Monitor for this step. Supports variable resolution (e.g., "Translating to {{ flowContext.targetLang }}..."). Falls back to the step name if not provided.

Optional steps to execute before showing the selection modal. Useful to prepare or transform input.

Prompt or title text shown at the top of the selection modal (supports variable resolution).

Optional title shown above additional info in the selection modal.

Optional descriptive text shown in the selection modal (supports variable resolution).

Enable keyboard navigation in the selection modal.

Input mappings from flow context to step inputs

Output mappings from step outputs to flow context

Shows an image picker modal allowing the user to pick one image from a provided list of image objects ({ url, previewUrl?, label? }). The selected image object is stored as an output. (Note: Aliases USER_IMAGE_PICKER are deprecated, use 'IMAGE_PICKER' instead)

Optional user-friendly label shown in the Flow Monitor for this step. Supports variable resolution (e.g., "Translating to {{ flowContext.targetLang }}..."). Falls back to the step name if not provided.

Prompt text shown in the image picker modal.

List of image objects to show. Each item should contain at least a url and optionally previewUrl and label.

Output mappings from step outputs to flow context

Show a progress bar with optional status text. Use position to place it and autoHide to auto-hide after completion.

Optional user-friendly label shown in the Flow Monitor for this step. Supports variable resolution (e.g., "Translating to {{ flowContext.targetLang }}..."). Falls back to the step name if not provided.

Status text to display next to the progress bar. Supports variable resolution.

Position of the progress bar. Common values: 'bottom', 'top'.

Whether the progress bar should auto-hide when completed.

Update an existing progress bar. Provide percentage (0-100) and optional status text.

Progress percentage (0-100). Can be a number or a string that resolves to a number.

Optional user-friendly label shown in the Flow Monitor for this step. Supports variable resolution (e.g., "Translating to {{ flowContext.targetLang }}..."). Falls back to the step name if not provided.

Status text to update on the progress bar. Supports variable resolution.

Hide any shown progress bar.

Optional user-friendly label shown in the Flow Monitor for this step. Supports variable resolution (e.g., "Translating to {{ flowContext.targetLang }}..."). Falls back to the step name if not provided.

Pause execution for a given delay (ms). Useful for pacing flows or waiting for external state changes.

Optional user-friendly label shown in the Flow Monitor for this step. Supports variable resolution (e.g., "Translating to {{ flowContext.targetLang }}..."). Falls back to the step name if not provided.

Delay in milliseconds to sleep. Defaults to 1000 ms.

Clears the current text selection and optionally repositions the cursor to the start or end of the selection. Available in Swing editor context only.

Optional user-friendly label shown in the Flow Monitor for this step. Supports variable resolution (e.g., "Translating to {{ flowContext.targetLang }}..."). Falls back to the step name if not provided.

Where to move the cursor after clearing selection. Use "anchorStart" to move to the start or "anchorEnd" to move to the end.

Requests a change of the view port size. Useful to resize the command palette view in Prime 8.

Target width in pixels. Can be a number or an expression resolved against the flowContext.

Target height in pixels. Can be a number or an expression resolved against the flowContext.

Optional user-friendly label shown in the Flow Monitor for this step. Supports variable resolution (e.g., "Translating to {{ flowContext.targetLang }}..."). Falls back to the step name if not provided.

Translate text using DeepL. Supports many optional DeepL parameters and returns the translated text as the default text output. (Note: Aliases SERVICE_DEEPL_TRANSLATE are deprecated, use 'DEEPL_TRANSLATE' instead)

Text to translate (can include templates). If omitted, the default text input is used.

Target language code for the translation (e.g. EN, DE, FR).

Optional user-friendly label shown in the Flow Monitor for this step. Supports variable resolution (e.g., "Translating to {{ flowContext.targetLang }}..."). Falls back to the step name if not provided.

Optional service name override. If not specified, defaults to 'DEEPL'.

Optional: replace xml:lang attributes in the translated output with this language code.

How to handle tags; default is "xml".

Language of the text to be translated. If this parameter is omitted, the API will attempt to detect the language of the text and translate it.

Additional context that can influence a translation but is not translated itself.

Sets whether the translation engine should first split the input into sentences.

Sets whether the translation engine should respect the original formatting, even if it would usually

Sets whether the translated text should lean towards formal or informal language.

Specify the glossary to use for the translation.

Disable the automatic detection of XML

Comma-separated list of XML tags which never split sentences.

Comma-separated list of XML tags which always cause splits.

Comma-separated list of XML tags that indicate text not to be translated.

Specifies which DeepL model should be used for translation.

When true, the response will include the billed_characters parameter.

Output mappings from step outputs to flow context

Rephrase/write text using DeepL Write endpoint. Returns the rephrased text in the default text output. (Note: Aliases SERVICE_DEEPL_WRITE are deprecated, use 'DEEPL_WRITE' instead)

Optional user-friendly label shown in the Flow Monitor for this step. Supports variable resolution (e.g., "Translating to {{ flowContext.targetLang }}..."). Falls back to the step name if not provided.

Optional service name override. If not specified, defaults to 'DEEPL_WRITE'.

Text to be rephrased or written. If omitted, default text input is used.

Optional target language for writing/rephrasing.

Optional writing style parameter.

Optional tone parameter for the writing API.

Output mappings from step outputs to flow context

Create or reuse an OpenAI assistant. If cfg.assistantId is provided the assistant will be retrieved; otherwise a new assistant is created. The created or retrieved assistant is written to the flow context.

Optional user-friendly label shown in the Flow Monitor for this step. Supports variable resolution (e.g., "Translating to {{ flowContext.targetLang }}..."). Falls back to the step name if not provided.

Optional service name override. If not specified, defaults to 'OPENAI_COMPLETION'.

If provided, load the assistant with this ID instead of creating a new one.

If true and storeIn is provided, try to reuse a persisted assistant id.

Storage location for persisting assistant id: page/session/browser.

Initial instructions / system prompt for the assistant (supports templates).

Model to use when creating the assistant

Output mappings from step outputs to flow context

Create or reuse an OpenAI thread. If cfg.reuse and cfg.storeIn are provided the thread id may be reused. The created thread is written to the flow context.

Optional user-friendly label shown in the Flow Monitor for this step. Supports variable resolution (e.g., "Translating to {{ flowContext.targetLang }}..."). Falls back to the step name if not provided.

Optional service name override. If not specified, defaults to 'OPENAI_COMPLETION'.

If true and storeIn provided, try to reuse a persisted thread id

Storage location for persisting thread id: page/session/browser.

Output mappings from step outputs to flow context

Delete a thread using its id. Provide the thread object or id via inputs.

Optional user-friendly label shown in the Flow Monitor for this step. Supports variable resolution (e.g., "Translating to {{ flowContext.targetLang }}..."). Falls back to the step name if not provided.

Optional service name override. If not specified, defaults to 'OPENAI_COMPLETION'.

Input mappings from flow context to step inputs

Create a vector store for thread file tools and optionally upload files. Writes created store to flow context and optional uploaded file list to an optional output.

Optional user-friendly label shown in the Flow Monitor for this step. Supports variable resolution (e.g., "Translating to {{ flowContext.targetLang }}..."). Falls back to the step name if not provided.

Optional service name override. If not specified, defaults to 'OPENAI_COMPLETION'.

Name of the vector store

If true replace existing stores instead of appending

Expiration configuration for the vector store

Input mappings from flow context to step inputs

Output mappings from step outputs to flow context

Send a message to a thread/assistant. The assistant and thread inputs are required. The message can be provided via cfg.instruction or via default text input. The assistant response text is stored in the default text output.

Optional user-friendly label shown in the Flow Monitor for this step. Supports variable resolution (e.g., "Translating to {{ flowContext.targetLang }}..."). Falls back to the step name if not provided.

Optional service name override. If not specified, defaults to 'OPENAI_COMPLETION'.

Optional message instruction; if omitted the default text input is used.

Input mappings from flow context to step inputs

Output mappings from step outputs to flow context

Chat/completion step that calls OpenAI (or Hub/Azure variants). Supports messages, images, audio, attachments, tools (functions), structured outputs (json_schema/json_object/list), reasoning and audio output. Many step-level configuration options are supported via cfg.*. (Note: Aliases SERVICE_OPENAI_COMPLETION are deprecated, use 'OPENAI_COMPLETION' instead)

Optional user-friendly label shown in the Flow Monitor for this step. Supports variable resolution (e.g., "Translating to {{ flowContext.targetLang }}..."). Falls back to the step name if not provided.

Optional service name override. If not specified, defaults to 'OPENAI_COMPLETION'.

Model id or deployment id to use

Id of a stored prompt configuration to reuse

The user prompt to send to the LLM (supports templates)

The system prompt to send to the LLM (supports templates)

Additional request options object (temperature, n, top_p, model, etc.)

Structured response format. Can be a string ("json_object" or "list") or an object with json_schema definition

Configuration object to request audio output (voice/format)

Optional tool selection policy (mapped to tool_choice in the model request)

Allow model/providers to emit multiple tool calls in one assistant turn (mapped to parallel_tool_calls)

Maximum number of tool iterations to perform

Behavior after a tool result (none|auto)

How to handle tool execution errors: "retry" allows LLM to retry failed tools (default), "fail" stops the flow immediately on any tool error, "once" prevents retrying the same tool after it fails once

How to handle function execution errors (deprecated, use toolErrorBehavior instead)

Array of tool/function descriptors for tool calling. Each descriptor can define template, steps, script, or scriptRef.

Array of function descriptors for tool calling (deprecated, use tools instead). Supports template, steps, script, or scriptRef.

Whether to reuse full flowContext when executing tool templates

Whether to reuse full flowContext when executing function templates (deprecated, use toolsReuseContext instead)

Optional safety identifier for the request

Array of prior messages to include in the conversation (overrides flowContext.messages)

Array or single image input(s) defined inline via cfg.images or cfg.image

Single inline image config

Single inline audio config

Array of audio inputs

Reasoning configuration object to pass to the API (for models supporting reasoning)

Image detail level for image inputs (low, high, auto)

Audio format for audio inputs (e.g., wav, mp3)

Array of generic attachments (images, audio, text) to include in the user message

'Enable dynamic tool discovery. When true, the agent can discover and activate tool providers on demand ' + 'via listAvailableTools and activateTools meta-tools, instead of requiring all tools to be listed explicitly. ' + 'Can be combined with builtinTools for aliases, MCP tools, and explicit provider configurations.'

List of built-in tool classes to enable for the AI model

Optional tool schema guidance (e.g., addTableFeedback columns or addMetricsFeedback metric keys and ranges)

Enable strict tool argument validation (reject missing/extra arguments)

Enable streaming mode for incremental response delivery. When enabled, tokens are delivered as they are generated rather than waiting for the full response. Streaming is automatically disabled for incompatible configurations (json_schema, list, audio output). Defaults to false.

Show streaming response in the Flow Execution Monitor. Defaults to true when stream is enabled. Set to false to use streaming for performance (faster time-to-first-token) without monitor visualization.

'List of skill names to load. Skills provide system instructions and reference documents. ' + 'Each skill is loaded from /SysConfig/ProActions/Skills/<skillName>/SKILL.md.'

Output mappings from step outputs to flow context

Chat/completion step that calls OpenAI (or Hub/Azure variants). Supports messages, images, audio, attachments, tools (functions), structured outputs (json_schema/json_object/list), reasoning and audio output. Many step-level configuration options are supported via cfg.*.

Optional user-friendly label shown in the Flow Monitor for this step. Supports variable resolution (e.g., "Translating to {{ flowContext.targetLang }}..."). Falls back to the step name if not provided.

Optional service name override. If not specified, defaults to 'HUB'.

Model id or deployment id to use

Id of a stored prompt configuration to reuse

The user prompt to send to the LLM (supports templates)

The system prompt to send to the LLM (supports templates)

Additional request options object (temperature, n, top_p, model, etc.)

Structured response format. Can be a string ("json_object" or "list") or an object with json_schema definition

Configuration object to request audio output (voice/format)

Optional tool selection policy (mapped to tool_choice in the model request)

Allow model/providers to emit multiple tool calls in one assistant turn (mapped to parallel_tool_calls)

Maximum number of tool iterations to perform

Behavior after a tool result (none|auto)

How to handle tool execution errors: "retry" allows LLM to retry failed tools (default), "fail" stops the flow immediately on any tool error, "once" prevents retrying the same tool after it fails once

How to handle function execution errors (deprecated, use toolErrorBehavior instead)

Array of tool/function descriptors for tool calling. Each descriptor can define template, steps, script, or scriptRef.

Array of function descriptors for tool calling (deprecated, use tools instead). Supports template, steps, script, or scriptRef.

Whether to reuse full flowContext when executing tool templates

Whether to reuse full flowContext when executing function templates (deprecated, use toolsReuseContext instead)

Optional safety identifier for the request

Array of prior messages to include in the conversation (overrides flowContext.messages)

Array or single image input(s) defined inline via cfg.images or cfg.image

Single inline image config

Single inline audio config

Array of audio inputs

Reasoning configuration object to pass to the API (for models supporting reasoning)

Image detail level for image inputs (low, high, auto)

Audio format for audio inputs (e.g., wav, mp3)

Array of generic attachments (images, audio, text) to include in the user message

'Enable dynamic tool discovery. When true, the agent can discover and activate tool providers on demand ' + 'via listAvailableTools and activateTools meta-tools, instead of requiring all tools to be listed explicitly. ' + 'Can be combined with builtinTools for aliases, MCP tools, and explicit provider configurations.'

List of built-in tool classes to enable for the AI model

Optional tool schema guidance (e.g., addTableFeedback columns or addMetricsFeedback metric keys and ranges)

Enable strict tool argument validation (reject missing/extra arguments)

Enable streaming mode for incremental response delivery. When enabled, tokens are delivered as they are generated rather than waiting for the full response. Streaming is automatically disabled for incompatible configurations (json_schema, list, audio output). Defaults to false.

Show streaming response in the Flow Execution Monitor. Defaults to true when stream is enabled. Set to false to use streaming for performance (faster time-to-first-token) without monitor visualization.

'List of skill names to load. Skills provide system instructions and reference documents. ' + 'Each skill is loaded from /SysConfig/ProActions/Skills/<skillName>/SKILL.md.'

Override the target used on service configuration level.

Output mappings from step outputs to flow context

Chat/completion step that calls OpenAI (or Hub/Azure variants). Supports messages, images, audio, attachments, tools (functions), structured outputs (json_schema/json_object/list), reasoning and audio output. Many step-level configuration options are supported via cfg.*. (Note: Aliases SERVICE_AZURE_OPENAI_COMPLETION are deprecated, use 'AZURE_OPENAI_COMPLETION' instead)

Optional user-friendly label shown in the Flow Monitor for this step. Supports variable resolution (e.g., "Translating to {{ flowContext.targetLang }}..."). Falls back to the step name if not provided.

Optional service name override. If not specified, defaults to 'AZURE_OPENAI_COMPLETION'.

Model id or deployment id to use

Id of a stored prompt configuration to reuse

The user prompt to send to the LLM (supports templates)

The system prompt to send to the LLM (supports templates)

Additional request options object (temperature, n, top_p, model, etc.)

Structured response format. Can be a string ("json_object" or "list") or an object with json_schema definition

Configuration object to request audio output (voice/format)

Optional tool selection policy (mapped to tool_choice in the model request)

Allow model/providers to emit multiple tool calls in one assistant turn (mapped to parallel_tool_calls)

Maximum number of tool iterations to perform

Behavior after a tool result (none|auto)

How to handle tool execution errors: "retry" allows LLM to retry failed tools (default), "fail" stops the flow immediately on any tool error, "once" prevents retrying the same tool after it fails once

How to handle function execution errors (deprecated, use toolErrorBehavior instead)

Array of tool/function descriptors for tool calling. Each descriptor can define template, steps, script, or scriptRef.

Array of function descriptors for tool calling (deprecated, use tools instead). Supports template, steps, script, or scriptRef.

Whether to reuse full flowContext when executing tool templates

Whether to reuse full flowContext when executing function templates (deprecated, use toolsReuseContext instead)

Optional safety identifier for the request

Array of prior messages to include in the conversation (overrides flowContext.messages)

Array or single image input(s) defined inline via cfg.images or cfg.image

Single inline image config

Single inline audio config

Array of audio inputs

Reasoning configuration object to pass to the API (for models supporting reasoning)

Image detail level for image inputs (low, high, auto)

Audio format for audio inputs (e.g., wav, mp3)

Array of generic attachments (images, audio, text) to include in the user message

'Enable dynamic tool discovery. When true, the agent can discover and activate tool providers on demand ' + 'via listAvailableTools and activateTools meta-tools, instead of requiring all tools to be listed explicitly. ' + 'Can be combined with builtinTools for aliases, MCP tools, and explicit provider configurations.'

List of built-in tool classes to enable for the AI model

Optional tool schema guidance (e.g., addTableFeedback columns or addMetricsFeedback metric keys and ranges)

Enable strict tool argument validation (reject missing/extra arguments)

Enable streaming mode for incremental response delivery. When enabled, tokens are delivered as they are generated rather than waiting for the full response. Streaming is automatically disabled for incompatible configurations (json_schema, list, audio output). Defaults to false.

Show streaming response in the Flow Execution Monitor. Defaults to true when stream is enabled. Set to false to use streaming for performance (faster time-to-first-token) without monitor visualization.

'List of skill names to load. Skills provide system instructions and reference documents. ' + 'Each skill is loaded from /SysConfig/ProActions/Skills/<skillName>/SKILL.md.'

Output mappings from step outputs to flow context

Upload an image to BrighterAI, poll for processing completion and download the anonymized image. The resulting anonymized image blob is written to the configured output. (Note: Aliases SERVICE_BRIGHTER_AI are deprecated, use 'BRIGHTER_AI' instead)

Optional user-friendly label shown in the Flow Monitor for this step. Supports variable resolution (e.g., "Translating to {{ flowContext.targetLang }}..."). Falls back to the step name if not provided.

Optional service name override. If not specified, defaults to 'BRIGHTER_AI'.

Service operation name (e.g. blur, dnat, mask). If not provided, blur is used.

Optional url parameters object passed to the BrighterAI upload endpoint.

Input mappings from flow context to step inputs

Output mappings from step outputs to flow context

Generate speech audio from text using ElevenLabs TTS. Text is taken from the default text input or a flow variable via in.

Optional user-friendly label shown in the Flow Monitor for this step. Supports variable resolution (e.g., "Translating to {{ flowContext.targetLang }}..."). Falls back to the step name if not provided.

Optional service name override. If not specified, defaults to 'ELEVENLABS'.

Override API key from service configuration

Override endpoint URL from service configuration

Voice identifier to use for synthesis

TTS model id

Optional voice settings object

Desired audio output format (e.g. mp3, wav, pcm)

Language code for the synthesis

Pronunciation dictionary locators (array or string) - resolved

Seed for deterministic generation

Optional previous_text context

Optional next_text context

Apply normalization to text

Apply language specific normalization

MIME type to use for the produced blob (e.g. audio/mpeg)

Input mappings from flow context to step inputs

Output mappings from step outputs to flow context

Transcribe audio using ElevenLabs STT. Provides the transcribed text and optional JSON result.

Optional user-friendly label shown in the Flow Monitor for this step. Supports variable resolution (e.g., "Translating to {{ flowContext.targetLang }}..."). Falls back to the step name if not provided.

Optional service name override. If not specified, defaults to 'ELEVENLABS'.

Override API key from service configuration

Override endpoint URL from service configuration

Model id to use for transcription - default: scribe_v1

Language code for transcription

Tag audio events option

Number of speakers to detect

Timestamps granularity

Enable diarization

Additional output formats

Input file format (e.g. wav, mp3)

URL to fetch input from cloud

Input mappings from flow context to step inputs

Output mappings from step outputs to flow context

Generate images using OpenAI image models (DALL·E and GPT-Image-1). Supports model, size, quality, count (n), response_format and other parameters. The generated images are stored in the configured output key or default imageList. (Note: Aliases SERVICE_OPENAI_IMAGE_GENERATION are deprecated, use 'OPENAI_IMAGE_GENERATION' instead)

Optional user-friendly label shown in the Flow Monitor for this step. Supports variable resolution (e.g., "Translating to {{ flowContext.targetLang }}..."). Falls back to the step name if not provided.

Optional service name override. If not specified, defaults to 'OPENAI_IMAGE_GENERATION'.

Image model to use (gpt-image-1, dall-e-3, dall-e-2)

The user prompt to send to the LLM (supports templates)

Desired image size (model-dependent)

Image quality setting

Number of images to generate (1-4; DALL·E3 supports 1)

Image style for DALL-E 3 (natural, vivid)

DALL·E response format: 'url' or 'b64_json' (not supported for gpt-image-1)

Output format for gpt-image-1 only (png|jpeg|webp)

Background setting for gpt-image-1 only (transparent)

Random seed for reproducible results (gpt-image-1 only)

User identifier for attribution and policy monitoring

Request timeout in milliseconds

Maximum number of retries for recoverable errors

FlowContext key where images will be stored (default: imageList)

Input mappings from flow context to step inputs

Output mappings from step outputs to flow context

ImageGeneration using ProActions Hub.

Optional user-friendly label shown in the Flow Monitor for this step. Supports variable resolution (e.g., "Translating to {{ flowContext.targetLang }}..."). Falls back to the step name if not provided.

Optional service name override. If not specified, defaults to 'HUB'.

Image model to use (gpt-image-1, dall-e-3, dall-e-2)

The user prompt to send to the LLM (supports templates)

Desired image size (model-dependent)

Image quality setting

Number of images to generate (1-4; DALL·E3 supports 1)

Image style for DALL-E 3 (natural, vivid)

DALL·E response format: 'url' or 'b64_json' (not supported for gpt-image-1)

Output format for gpt-image-1 only (png|jpeg|webp)

Background setting for gpt-image-1 only (transparent)

Random seed for reproducible results (gpt-image-1 only)

User identifier for attribution and policy monitoring

Request timeout in milliseconds

Maximum number of retries for recoverable errors

FlowContext key where images will be stored (default: imageList)

Override the target used on service configuration level.

Input mappings from flow context to step inputs

Output mappings from step outputs to flow context

ImageGeneration using Azure OpenAI. (Note: Aliases SERVICE_AZURE_OPENAI_IMAGE_GENERATION are deprecated, use 'AZURE_OPENAI_IMAGE_GENERATION' instead)

Optional user-friendly label shown in the Flow Monitor for this step. Supports variable resolution (e.g., "Translating to {{ flowContext.targetLang }}..."). Falls back to the step name if not provided.

Optional service name override. If not specified, defaults to 'AZURE_OPENAI_IMAGE_GENERATION'.

Image model to use (gpt-image-1, dall-e-3, dall-e-2)

The user prompt to send to the LLM (supports templates)

Desired image size (model-dependent)

Image quality setting

Number of images to generate (1-4; DALL·E3 supports 1)

Image style for DALL-E 3 (natural, vivid)

DALL·E response format: 'url' or 'b64_json' (not supported for gpt-image-1)

Output format for gpt-image-1 only (png|jpeg|webp)

Background setting for gpt-image-1 only (transparent)

Random seed for reproducible results (gpt-image-1 only)

User identifier for attribution and policy monitoring

Request timeout in milliseconds

Maximum number of retries for recoverable errors

FlowContext key where images will be stored (default: imageList)

Input mappings from flow context to step inputs

Output mappings from step outputs to flow context

Generate speech audio using the OpenAI Speech API (or corresponding Hub/Azure variants). Reads text from the default text input and returns an audio blob. (Note: Aliases SERVICE_OPENAI_SPEECH are deprecated, use 'OPENAI_SPEECH' instead)

Optional user-friendly label shown in the Flow Monitor for this step. Supports variable resolution (e.g., "Translating to {{ flowContext.targetLang }}..."). Falls back to the step name if not provided.

Optional service name override. If not specified, defaults to 'OPENAI_SPEECH'.

Model id to use for TTS (e.g. tts-1)

Voice id to use for synthesis

Response audio format (e.g. mp3, wav, pcm)

Playback speed multiplier (e.g. 1.0)

Optional MIME type for the produced blob (e.g. audio/mpeg)

Input mappings from flow context to step inputs

Output mappings from step outputs to flow context

ProActionsHub-compatible speech generation step. Inherits behavior from OPENAI_SPEECH.

Optional user-friendly label shown in the Flow Monitor for this step. Supports variable resolution (e.g., "Translating to {{ flowContext.targetLang }}..."). Falls back to the step name if not provided.

Optional service name override. If not specified, defaults to 'HUB'.

Model id to use for TTS (e.g. tts-1)

Voice id to use for synthesis

Response audio format (e.g. mp3, wav, pcm)

Playback speed multiplier (e.g. 1.0)

Optional MIME type for the produced blob (e.g. audio/mpeg)

Override the target used on service configuration level.

Input mappings from flow context to step inputs

Output mappings from step outputs to flow context

Azure OpenAI speech step alias. (Note: Aliases SERVICE_AZURE_OPENAI_SPEECH are deprecated, use 'AZURE_OPENAI_SPEECH' instead)

Optional user-friendly label shown in the Flow Monitor for this step. Supports variable resolution (e.g., "Translating to {{ flowContext.targetLang }}..."). Falls back to the step name if not provided.

Optional service name override. If not specified, defaults to 'AZURE_OPENAI_SPEECH'.

Model id to use for TTS (e.g. tts-1)

Voice id to use for synthesis

Response audio format (e.g. mp3, wav, pcm)

Playback speed multiplier (e.g. 1.0)

Optional MIME type for the produced blob (e.g. audio/mpeg)

Input mappings from flow context to step inputs

Output mappings from step outputs to flow context

Transcribe audio using OpenAI (or Azure/Hub variants). Reads an audio file/blob from inputs and writes the transcription text and optional segment list. (Note: Aliases SERVICE_OPENAI_TRANSCRIPTION are deprecated, use 'OPENAI_TRANSCRIPTION' instead)

Optional user-friendly label shown in the Flow Monitor for this step. Supports variable resolution (e.g., "Translating to {{ flowContext.targetLang }}..."). Falls back to the step name if not provided.

Optional service name override. If not specified, defaults to 'OPENAI_TRANSCRIPTION'.

Optional prompt/instruction to bias transcription.

Optional language code to hint at language for transcription.

Optional temperature / randomness parameter for transcription model.

Optional model override (e.g. 'whisper-1').

Input mappings from flow context to step inputs

Output mappings from step outputs to flow context

Hub-compatible transcription step (delegates to hub service). Writes transcription text and optional segments list. (Note: Aliases SERVICE_OPENAI_TRANSCRIPTION are deprecated, use 'HUB_TRANSCRIPTION' instead)

Optional user-friendly label shown in the Flow Monitor for this step. Supports variable resolution (e.g., "Translating to {{ flowContext.targetLang }}..."). Falls back to the step name if not provided.

Optional service name override. If not specified, defaults to 'HUB'.

Optional prompt/instruction to bias transcription.

Optional language code to hint at language for transcription.

Optional temperature / randomness parameter for transcription model.

Optional model override (e.g. 'whisper-1').

Input mappings from flow context to step inputs

Output mappings from step outputs to flow context

Azure OpenAI transcription alias (delegates to the OpenAI transcription implementation). (Note: Aliases SERVICE_AZURE_OPENAI_TRANSCRIPTION are deprecated, use 'AZURE_OPENAI_TRANSCRIPTION' instead)

Optional user-friendly label shown in the Flow Monitor for this step. Supports variable resolution (e.g., "Translating to {{ flowContext.targetLang }}..."). Falls back to the step name if not provided.

Optional service name override. If not specified, defaults to 'AZURE_OPENAI_TRANSCRIPTION'.

Optional prompt/instruction to bias transcription.

Optional language code to hint at language for transcription.

Optional temperature / randomness parameter for transcription model.

Optional model override (e.g. 'whisper-1').

Input mappings from flow context to step inputs

Output mappings from step outputs to flow context

Upscale an image using StabilityAI upscale endpoint. Uploads a provided image blob and returns the upscaled image blob. (Note: Aliases SERVICE_STABILITY_AI_UPSCALE are deprecated, use 'STABILITY_AI_UPSCALE' instead)

Optional user-friendly label shown in the Flow Monitor for this step. Supports variable resolution (e.g., "Translating to {{ flowContext.targetLang }}..."). Falls back to the step name if not provided.

Optional service name override. If not specified, defaults to 'STABILITY_AI'.

Optional prompt describing desired changes to the image

Desired output format (e.g. jpeg, png)

Input mappings from flow context to step inputs

Output mappings from step outputs to flow context

Outpaint an image region using StabilityAI. Supports coordinates (left,right,up,down) to define the outpainting area. (Note: Aliases SERVICE_STABILITY_AI_OUTPAINT are deprecated, use 'STABILITY_AI_OUTPAINT' instead)

Optional user-friendly label shown in the Flow Monitor for this step. Supports variable resolution (e.g., "Translating to {{ flowContext.targetLang }}..."). Falls back to the step name if not provided.

Optional service name override. If not specified, defaults to 'STABILITY_AI'.

Pixels to the left (default 0)

Pixels to the right (default 0)

Pixels to the top (default 0)

Pixels to the bottom (default 0)

Input mappings from flow context to step inputs

Output mappings from step outputs to flow context

Search and replace regions or objects within an image using StabilityAI. Provide a prompt and a search_prompt to locate and replace visual elements. (Note: Aliases SERVICE_STABILITY_AI_SEARCH_AND_REPLACE are deprecated, use 'STABILITY_AI_SEARCH_AND_REPLACE' instead)

Optional user-friendly label shown in the Flow Monitor for this step. Supports variable resolution (e.g., "Translating to {{ flowContext.targetLang }}..."). Falls back to the step name if not provided.

Optional service name override. If not specified, defaults to 'STABILITY_AI'.

Prompt describing replacement target or desired result

Prompt describing what to search for in the image

Input mappings from flow context to step inputs

Output mappings from step outputs to flow context

Search and recolor elements in an image using StabilityAI. Provide select_prompt and prompt to find and recolor elements. (Note: Aliases SERVICE_STABILITY_AI_SEARCH_AND_RECOLOR are deprecated, use 'STABILITY_AI_SEARCH_AND_RECOLOR' instead)

Optional user-friendly label shown in the Flow Monitor for this step. Supports variable resolution (e.g., "Translating to {{ flowContext.targetLang }}..."). Falls back to the step name if not provided.

Optional service name override. If not specified, defaults to 'STABILITY_AI'.

Prompt to select elements to recolor

Prompt describing the recoloring operation

Input mappings from flow context to step inputs

Output mappings from step outputs to flow context

Invoke a tool on an MCP server via the Hub.

The name of the MCP server.

The name of the tool to invoke.

Optional user-friendly label shown in the Flow Monitor for this step. Supports variable resolution (e.g., "Translating to {{ flowContext.targetLang }}..."). Falls back to the step name if not provided.

Optional service name override. If not specified, defaults to 'HUB'.

The arguments to pass to the tool.

Output mappings from step outputs to flow context

List available tools from an MCP server via the Hub.

Optional user-friendly label shown in the Flow Monitor for this step. Supports variable resolution (e.g., "Translating to {{ flowContext.targetLang }}..."). Falls back to the step name if not provided.

Optional service name override. If not specified, defaults to 'HUB'.

The name of the MCP server to list tools from.

Output mappings from step outputs to flow context

Fetch object content from EDAPI and return it as blob or configured outputs. The content id can be provided via cfg.contentId or via a text input.

Optional user-friendly label shown in the Flow Monitor for this step. Supports variable resolution (e.g., "Translating to {{ flowContext.targetLang }}..."). Falls back to the step name if not provided.

Explicit content id to fetch. If omitted, the step will try to read the id from the default text input.

Format of the content to fetch.

Input mappings from flow context to step inputs

Output mappings from step outputs to flow context

Upload an image to EDAPI by providing an image object (url or blob). Returns the created object metadata as output. (Note: Aliases CTX_UPLOAD_IMAGE are deprecated, use 'UPLOAD_IMAGE' instead)

Optional user-friendly label shown in the Flow Monitor for this step. Supports variable resolution (e.g., "Translating to {{ flowContext.targetLang }}..."). Falls back to the step name if not provided.

Type of object to create (default: Image).

Creation mode, e.g. AUTO_RENAME.

Input mappings from flow context to step inputs

Output mappings from step outputs to flow context

Update binary content of an existing object in EDAPI using the provided blob. Returns updated object metadata. (Note: Aliases CTX_UPDATE_BINARY_CONTENT are deprecated, use 'UPDATE_BINARY_CONTENT' instead)

Optional user-friendly label shown in the Flow Monitor for this step. Supports variable resolution (e.g., "Translating to {{ flowContext.targetLang }}..."). Falls back to the step name if not provided.

Input mappings from flow context to step inputs

Output mappings from step outputs to flow context

Upload content to EDAPI by providing a blob or specifying parameters (filename, type). Returns created object metadata.

Optional user-friendly label shown in the Flow Monitor for this step. Supports variable resolution (e.g., "Translating to {{ flowContext.targetLang }}..."). Falls back to the step name if not provided.

Obtions data for the UPLOAD

Filename to use for the uploaded object

BaseType - Used to identify upload location when using basefolder configuration.

Object type to create (e.g. File, Image)

Creation mode (e.g. AUTO_RENAME)

Input mappings from flow context to step inputs

Output mappings from step outputs to flow context

Initiate OAuth authorization flow for YouTube (requires ProActions-Hub). Opens the auth URL when available unless cfg.omitOpenAuth is true.

Optional user-friendly label shown in the Flow Monitor for this step. Supports variable resolution (e.g., "Translating to {{ flowContext.targetLang }}..."). Falls back to the step name if not provided.

Optional service name override. If not specified, defaults to 'HUB'.

If true, do not automatically open the authorization URL in a new window; return the URL in the response instead.

Optional account identifier used by the Hub API.

Output mappings from step outputs to flow context

Check the status of a previously initiated YouTube OAuth authorization flow.

Optional user-friendly label shown in the Flow Monitor for this step. Supports variable resolution (e.g., "Translating to {{ flowContext.targetLang }}..."). Falls back to the step name if not provided.

Optional service name override. If not specified, defaults to 'HUB'.

Optional account identifier used by the Hub API.

Output mappings from step outputs to flow context

Log out the configured Hub YouTube account. Uses POST to logout and returns the resulting object.

Optional user-friendly label shown in the Flow Monitor for this step. Supports variable resolution (e.g., "Translating to {{ flowContext.targetLang }}..."). Falls back to the step name if not provided.

Optional service name override. If not specified, defaults to 'HUB'.

Account identifier to log out (resolved).

Output mappings from step outputs to flow context

Upload a video to YouTube via the Hub service. Supports video file, optional thumbnail, metadata fields (title, description, privacyStatus, categoryId), tags and additionalMetadata. Progress can be shown when cfg.updateProgress is true.

Optional user-friendly label shown in the Flow Monitor for this step. Supports variable resolution (e.g., "Translating to {{ flowContext.targetLang }}..."). Falls back to the step name if not provided.

Optional service name override. If not specified, defaults to 'HUB'.

Video title

Video description

Privacy status: public, unlisted, or private

YouTube category id

Array of tags to set on the uploaded video

Optional object with additional metadata to attach

If true, progress will be reported to ProgressBar during upload

Account identifier for the Hub target (resolved)

Input mappings from flow context to step inputs

Output mappings from step outputs to flow context

Extract textual content from a URL using the ProActions-Hub extraction tool. The URL can be provided via the default text input or the in/url configuration.

Optional user-friendly label shown in the Flow Monitor for this step. Supports variable resolution (e.g., "Translating to {{ flowContext.targetLang }}..."). Falls back to the step name if not provided.

Optional service name override. If not specified, defaults to 'HUB'.

Input mappings from flow context to step inputs

Output mappings from step outputs to flow context

Generic REST step. Builds a Request from the given configuration (url, method, headers, parameters, body or formData) and executes it using the host HTTP helper. Results are written to cfg.outputs when provided or to the default text output. (Note: Aliases SERVICE_REST are deprecated, use 'REST' instead)

The URL to call. Supports template expressions and variable resolution.

Optional user-friendly label shown in the Flow Monitor for this step. Supports variable resolution (e.g., "Translating to {{ flowContext.targetLang }}..."). Falls back to the step name if not provided.

HTTP method to use (GET, POST, PUT, DELETE, ...). Default: GET.

Optional headers object. Values will be resolved against the flow context.

Optional query parameters object. Values will be resolved against the flow context.

Optional request body. May be a string or object. Objects are JSON-stringified by the step.

Optional formData object. Supports File/Blob entries and arrays. Values are resolved against the flow context.

Optional low-level RequestInit overrides passed to the fetch helper.

Output mappings from step outputs to flow context