Master the Cotonity visual canvas — adding and connecting nodes, mapping data between steps, saving drafts, and organizing your agent graph for long-term maintainability.
Opening the visual editor
Navigate to Agents in the left sidebar, then click New Agent. You will be placed on a blank canvas with a Start node already present. The canvas supports pan (middle-click drag or space-bar drag), zoom (scroll wheel or the zoom controls in the toolbar), and multi-select (shift-click or drag a selection box). Each node on the canvas represents one step in your agent's execution. Nodes are connected by edges that define the flow of data from one step to the next. You can save at any time with Cmd+S (Mac) or Ctrl+S (Windows/Linux); Cotonity auto-saves every 30 seconds as well.
Adding and configuring nodes
Click the + button that appears when you hover over any edge or node connector to open the node picker. Nodes are grouped into four categories: Triggers (how the agent starts), Actions (what the agent does), Logic (conditions and loops), and AI (LLM calls, prompt templates, and memory operations). After adding a node, click it to open its configuration panel on the right. Most nodes require you to map input fields from previous steps using the data-picker, which shows a tree of all available outputs from upstream nodes. Required fields are marked with a red asterisk; the editor will warn you before saving if any required fields are unmapped.
Connecting nodes and mapping data
Drag from the output handle (right side) of one node to the input handle (left side) of another to create an edge. You can create branching flows by connecting one node's output to multiple downstream nodes; by default all branches run in parallel. To make branches conditional, insert a Branch node and configure its conditions before connecting each outgoing edge to its respective path. Data from any upstream node is available in downstream nodes via the expression syntax `{{steps.nodeName.output.fieldName}}`. The data-picker helps you discover available fields without memorizing the syntax.
Saving, naming, and organizing your agent
Give your agent a descriptive name that reflects its purpose — for example, 'Enrich new lead from CRM' rather than 'Agent 1'. You can also add a description, assign it to a folder, and tag it with labels for easier filtering. Once you are satisfied with the canvas layout, click Save Draft to store the current state without publishing. Draft agents are only visible to you and cannot be triggered externally. When you are ready to test the full flow, use the Run in Sandbox button, which executes the agent with real logic but isolated from your production data and external side-effects.