Quickstart

1. Authorize the CLI

   Run the initializer from your application repository. It opens a browser for human authorization, creates an API key, writes it to .env.local, and installs the SDK.

   bash

   1npx @inkibra/voice-cli init

2. Create a short-lived session token

   Call the session endpoint from server-side code only.

   ts

   12345678910111213const response = await fetch(
     "https://inkibra-voice.inkibra.workers.dev/v1/sessions",
     {
       method: "POST",
       headers: {
         authorization: `Bearer ${process.env.INKIBRA_API_KEY}`,
         "content-type": "application/json",
       },
       body: JSON.stringify({ voice: "ara" }),
     },
   );
   
   const { token } = await response.json();

3. Connect in the browser

   ts

   1234567891011import LiveVoice from "@inkibra/voice-sdk";
   
   const session = await LiveVoice.connect({
     token,
     prompt: "Help the user complete a safety inspection.",
     conversation: {
       starts: "assistant",
       prompt: "Be concise. Clarify uncertain details before continuing.",
     },
     tools: {},
   });

4. Subscribe to events

   ts

   12345session.on("transcript.instantaneous", ({ text }) => updateOptimisticUI(text));
   session.on("transcript.canonical", ({ turns }) => saveConversationView(turns));
   session.on("floor", ({ owner, assistantAudible }) => updateVoiceUI(owner, assistantAudible));
   session.on("phase", ({ phase }) => updateWorkflow(phase));
   session.on("error", console.error);

5. Clean up

   ts

   12session.stop();
   session.dispose();