Visual workflow builder

What this accomplishes

The node graph editor in Miton’s Automation: triggers, actions, control flow, output nodes, the connection model, the per-node configuration, the integration node types, the keyboard shortcuts.

The visual workflow builder is the node graph editor at the core of the Automation. The builder is a drag-and-drop surface with a node graph, a node palette, a per-node inspector, and a connection model. The builder is the right shape for users who want to automate recurring work without writing code.

This guide covers the visual workflow builder. See the Automation overview for presets; the scheduler is in local scheduler; webhooks in webhook receiver; run history in run inspector and replay; safe testing in dry-run mode; integrations in integration nodes.

Current availability

This page documents behaviour verified in Miton development builds. Miton is still pre-release and is not publicly downloadable; invited beta users will receive a signed desktop build with automatic updates.

Before you start

• A Miton development build, an active project and Automation open.
• A connected model plus permission for every file, network or scheduled action in the workflow.

The node graph

The node graph is the panel showing the workflow you’re editing — a directed graph you build by dragging nodes onto the canvas and connecting one node’s output to another’s input.

The graph supports:

• Pan and zoom with the trackpad, mouse, or mouse wheel.
• Drag and drop — drag a node from the palette onto the graph, or drag an existing node to reposition it.
• Connect — drag from one node’s output handle to another node’s input handle. Branch, output, and approval nodes expose more than one named output handle (for example a branch’s true / false, or an approval’s approved / rejected).
• Delete — select a node and press Delete.
• Undo / redo — Cmd+Z / Cmd+Shift+Z.

The graph is per workflow and is preserved across sessions.

The node palette

The palette lists the node types you can add:

• Trigger — starts a run: manual, scheduled (cron), or webhook.
• Input — a typed workflow input (text, number, or JSON) with an optional default.
• LLM Call — a model call with its own provider, model, system and user prompt, temperature, and max tokens.
• Tool — invokes one of the agent’s tools by name against a templated input.
• Branch — evaluates a condition and routes to a true or false output.
• Loop — repeats while a condition holds, up to a maximum number of iterations.
• Merge — combines multiple incoming branches: take the first to arrive, wait for all, or concatenate.
• Code — runs a short JavaScript or Python snippet.
• Subagent — dispatches a task to a named sub-agent.
• Advance Goal — marks a project Goal or task complete.
• Write Memory — saves content to project memory.
• Output — the run’s result, formatted as text, JSON, or Markdown.
• HTTP, RSS, Email, File watcher, Git event, Approval — see integration nodes.

The connection model

Each connection is a directed edge from one node’s output handle to another’s input handle. Most node kinds have a single default input and output; branch, output, and approval expose the extra named handles noted above, and the workflow follows whichever one fires.

The per-node configuration

Selecting a node opens its inspector on the right, with the fields specific to that node kind — a schedule node’s cron expression, an LLM Call’s prompt and model, an HTTP node’s method and URL, and so on. Field values can be a constant or reference an earlier node’s output.

The keyboard shortcuts

• Cmd+K then T / A / C / O — add a trigger / action / control / output node.
• Cmd+Shift+R — open the run inspector.
• Cmd+Shift+D — open the dry-run dialog.

The full keyboard shortcut list, including Automation’s other chords, is in the reference keybindings.

What the visual workflow builder does not cover

• Real-time collaboration on the workflow — a workflow is single-user; there’s no live co-editing of the same graph.
• Version control for the workflow beyond the run history — the workflow lives in the project; use git for history on its underlying file the same way you would for any other project file.

What success looks like

You can drag nodes onto the graph, wire them together, configure each one, and run the workflow — without writing code — and each node’s inspector shows exactly the fields relevant to that node kind.

Common failures and recovery

• Inspect the failed run and the exact node output before replaying it.
• Confirm credentials, permissions and test fixtures belong to the active project.
• Use a dry run to isolate a misconfigured node before starting the complete workflow again.

Related next action

Open the Automation overview to connect this task to the rest of the workflow.