Skip to main content

Quickstart

Install the module, wire the Scan → Render loop into your host, and run the bundled example.

Install​

go get github.com/joestump-agent/a2tea

Wire up a host​

Scan the reply into parts, render each part's prose, and hand its messages to Render. Render returns an embeddable model — it never owns the terminal.

parts, err := a2tea.Scan(reply)
if err != nil {
    // not valid A2UI — render reply as plain text
}
for _, p := range parts {
    if p.Text != "" {
        renderProse(p.Text)
    }
    if len(p.Messages) > 0 {
        model, err := a2tea.Render(p.Messages)
        if err != nil {
            continue
        }
        draw(model)
    }
}
tip

Use a2tea.Contains(reply) as a cheap gate before taking the Scan path at all.

Run the example​

A runnable version of the whole flow lives in examples/hello. It wraps a single surface with Standalone so it runs on its own.

go run ./examples/hello
╭──────────────────────────╮
│ Hello from A2UI          │
│ rendered by a2tea        │
│ ──────────────────────── │
│ [ OK ]                   │
╰──────────────────────────╯
tab cycle • enter select • q quit