Skip to main content
Portal Agent executes actions on your canvas by calling named tools — each with a defined purpose and set of parameters. When you send a prompt, the agent chooses which tools to call, in what order, and with what arguments. Understanding the available tools helps you write more precise prompts and predict what the agent will do.

​
Brainstorming canvas tools

These tools operate on the infinite brainstorming canvas. They are available whenever you are in Brainstorming mode.
Creates a new card on the brainstorming canvas. The agent positions cards intelligently by default — you do not need to specify coordinates.
ParameterTypeRequiredDescription
contentstringYesText content for the card
widthnumberNoCard width in pixels (default: 200)
heightnumberNoCard height in pixels (default: 150)
typestringNotext, image, or ai-generated
backgroundColorstringNoCard color as a hex value
Example prompt: Create five cards with the core pillars of UX design
Creates a card and immediately populates it with meaningful content on a given topic. Use this when you want the agent to generate the card text itself rather than providing it explicitly.
ParameterTypeRequiredDescription
topicstringYesSubject for the card (e.g., “climate change”, “SWOT analysis”)
contentstringYesThe detailed content to populate the card with
titlestringNoOptional title; generated from topic if omitted
backgroundColorstringNoCard color as a hex value
Example prompt: Add a card about the benefits of async communication for remote teams
Edits the text content of an existing card by its ID.
ParameterTypeRequiredDescription
cardIdstringYesID of the card to edit
contentstringYesNew text content
Example prompt: Update the blue card to say "Q3 2025 targets"
Moves a card to an explicit canvas position.
ParameterTypeRequiredDescription
cardIdstringYesID of the card to move
xnumberYesTarget X coordinate in world space
ynumberYesTarget Y coordinate in world space
Example prompt: Rearrange the cards so the "Research" card is at the top
Positions a card relative to another card using spatial relationships like “below”, “to the right of”, or “centered above”. Prefer this over move_brainstorm_card when you want to express layout intent in natural language.
ParameterTypeRequiredDescription
cardIdstringYesID of the card to position
referenceCardIdstringYesID of the card to position relative to
sidestringYestop, bottom, left, or right
alignstringYesstart, center, or end
sideOffsetnumberNoDistance from the reference card (default: 10)
alignOffsetnumberNoOffset along the alignment axis (default: 0)
intentstringYesHuman-readable positioning description
Example prompt: Place the "Execution" card directly below the "Strategy" card
Draws a line or arrow connector between two cards. You can identify cards by their ID or by a description like “the red card” or “the first card”.
ParameterTypeRequiredDescription
fromCardIdstringNoSource card ID (if known)
toCardIdstringNoTarget card ID (if known)
fromDescriptionstringNoDescription of the source card (e.g., “red card”)
toDescriptionstringNoDescription of the target card
typestringNoline or arrow (default: arrow)
colorstringNoConnector color as a hex value
strokeWidthnumberNoLine thickness (default: 2)
Example prompt: Connect the "Problem" card to the "Solution" card with an arrow
Updates or removes an existing connector between cards.
ParameterTypeRequiredDescription
connectionIdstringYesID of the connection to modify
actionstringYesupdate or remove
typestringNoNew connector type (line or arrow)
colorstringNoNew connector color
strokeWidthnumberNoNew line thickness
Example prompt: Remove all connectors between the idea cards
Generates a structured layout template on the canvas from a central topic and list of subtopics.
ParameterTypeRequiredDescription
templateTypestringYesmindmap, wordweb, or flowchart
centralTopicstringYesThe root topic of the template
subtopicsarrayYes2–8 branch topics as strings
useAIGenerationbooleanNoLet AI expand the template with additional content
Example prompt: Create a mind map for a product launch with subtopics for marketing, design, engineering, and sales
Creates a named container area on the canvas with an internal layout for organizing cards placed inside it.
ParameterTypeRequiredDescription
namestringNoSection label shown in the section tab
xnumberNoWorld-space center X position
ynumberNoWorld-space center Y position
widthnumberNoSection width (default: 1200)
heightnumberNoSection height (default: 800)
layoutTypestringNorow, column, waterfall, grid, or bento
strokeColorstringNoSection border color
Example prompt: Create a "Research" section and a "Concepts" section side by side

​
Shape and drawing tools

Creates a single geometric shape at an auto-positioned or specified location. For complex objects (vehicles, figures, buildings), call this multiple times or use create_smart_shape instead.
ParameterTypeRequiredDescription
intentstringNoHuman-readable description of what the shape represents
typestringNocircle, rectangle, triangle, square, star, or heart
xnumberNoX position (omit for auto-positioning)
ynumberNoY position (omit for auto-positioning)
sizenumberNoShape size in pixels
fillstringNoFill color as a hex value
widthnumberNoWidth override (useful for rectangles)
heightnumberNoHeight override
rotationnumberNoRotation in degrees, around the center
Example prompt: Draw a large red circle in the center of the canvas
Creates a multi-component object by decomposing it into properly proportioned shapes — useful for recognizable objects like animals, vehicles, or buildings.
ParameterTypeRequiredDescription
objectstringYesWhat to create (e.g., “dog”, “spaceship”, “coffee cup”)
sizeMultipliernumberNoScale the whole object up or down (default: 1.0)
Example prompt: Draw a simple cat on the canvas
Creates multiple shapes in a single call to form a complex object. More reliable than sequential create_shape calls when building multi-part objects.
ParameterTypeRequiredDescription
objectTypestringYesType of object (e.g., “house”, “car”, “robot”)
shapesarrayYesArray of shape definitions with type, intent, color, size, and optional relativePosition
Example prompt: Build a simple house from shapes
Resizes or rotates an existing shape by ID.
ParameterTypeRequiredDescription
shapeIdstringYesID of the shape to resize
widthnumberNoNew width in pixels
heightnumberNoNew height in pixels
scalenumberNoUniform scale factor (1.0 = no change)
scaleXnumberNoHorizontal scale factor
scaleYnumberNoVertical scale factor
rotationnumberNoNew rotation in degrees
Example prompt: Make the circle twice as large
Renders a freehand sketch as a series of connected points.
ParameterTypeRequiredDescription
pointsarrayYesArray of {x, y} coordinate objects
colorstringNoStroke color (default: #000000)
strokeWidthnumberNoLine thickness (default: 3)
Example prompt: Sketch a rough arrow pointing right
Adds a floating text label at an explicit position on the canvas.
ParameterTypeRequiredDescription
textstringYesText content
xnumberYesX position in world space
ynumberYesY position in world space
fontSizenumberNoFont size in pixels (default: 16)
colorstringNoText color (default: #000000)
Example prompt: Add a heading "User Journey" at the top of the canvas

​
Workflow and utility tools

Executes a high-level creative workflow that plans and places a cohesive set of cards on the canvas using a deterministic layout strategy.
ParameterTypeRequiredDescription
promptstringYesHigh-level intent describing the topic and goals
layoutStrategystringNobento, waterfall, grid, or radial (default: bento)
workflowModestringNoresearch_board or moodboard_radial
bentoVariantstringNoBento composition pattern (default: auto)
maxObjectsnumberNoMaximum number of objects to create (4–20, default: 10)
Example prompt: Build a complete research board on climate change solutions
Takes a screenshot of the current canvas for the agent to visually analyze — useful before creating new content so the agent understands the existing layout.
ParameterTypeRequiredDescription
analysis_purposestringNolayout_analysis, content_placement, visual_context, or quality_check
Example prompt: Look at what's on the canvas and suggest how to improve the layout
Reads the current canvas state — all cards, their positions, sizes, and IDs — without taking a screenshot.
ParameterTypeRequiredDescription
intentstringNoWhy you need the canvas info
Example prompt: List all the cards currently on the canvas
Switches the workspace to a different mode.
ParameterTypeRequiredDescription
modestringYesbrainstorming, design, or compose
Example prompt: Switch to Compose mode and build a landing page
Exports the current canvas content to a file.
ParameterTypeRequiredDescription
formatstringNopng, jpg, svg, pdf, or json (default: png)
includeBackgroundbooleanNoInclude the canvas background (default: true)
Example prompt: Export the canvas as a PNG

​
Example prompts that combine tools

Create a mind map for a mobile app startup, then connect all branches back to the central idea

Add three cards for "Discovery", "Design", and "Launch", then draw arrows between them in sequence

Read the canvas, then reorganize all the cards into a grid layout with sections

Draw a simple robot using shapes, then add a label below it
You don’t need to name tools in your prompt. The agent selects the right tools automatically based on what you describe. Naming tools explicitly (e.g., “use create_brainstorm_card”) is only necessary when you need to override the agent’s default behavior.