Stop guessing what your agent did

A workbench, not a log viewer

Everything you'd otherwise wire up by hand — request builder, live stream, replay, diffing — in one place that speaks agent, not HTTP.

When your agent delegates, you can still see everything

Most tools flatten a spawned subagent into a single opaque tool call. Langport reconstructs the tree from the event stream — parent run, child runs, and every hop below — so a nested delegation reads like the conversation it actually was.

If it streams events, it works

Langport speaks plain SSE and NDJSON. Emit whatever shape you already emit — it maps tool calls, deltas, and parent ids on the way in. Point it at your agent and hit send.

// examples/agui/main.go — a Gantry agent over AG-UI.

func newHandler(llm gantry.LLMClient, opts ...agui.Option) (http.Handler, error) {
    agent, err := gantry.NewAgent(gantry.WithLLM(llm))
    if err != nil {
        return nil, err
    }
    // ask_user is a client-side tool: the run suspends,
    // the client answers, the run resumes.
    if err := agent.With(tool.Client(ask.Definition())); err != nil {
        return nil, err
    }
    return agui.Handler(agent, opts...), nil
}

http.Handle("/agui", handler)
// → point Langport at localhost:8080/agui

Logging tells you what happened. Langport lets you do it again.