TypeScriptIntermediate

Gemini Agora Voice Agents for Next.js

Build a Gemini-powered voice agent with Next.js and Agora.

Recipe prompt

Paste into Cursor, Claude Code, v0, or your coding agent

Use with your coding agent

You are implementing the "Gemini Agora Voice Agents for Next.js" recipe in this project.

Read the recipe markdown first:
https://raw.githubusercontent.com/AgoraIO-Community/Gemini-Agora-Voice-Agents-NextJS/main/docs/ai/RECIPE.md

Use the source repository for cross-reference:
https://github.com/AgoraIO-Community/Gemini-Agora-Voice-Agents-NextJS

Build this recipe into the user's app using the markdown as the implementation guide. Inspect related source files through the repository links when the recipe points to them. Ask before installing new dependencies.

Recipe

Rendered from the configured recipe markdown.

Raw

Recipe metadata
Field	Value
`recipe_version`	`0.1.0`
`recipe_status`	`stable`
`extension_points`	`api.routes` `prompts.system` `pipeline.providers` `ui.conversation`
`invariants`	`baseline.official-sample` `tokens.rtc-rtm` `lifecycle.strict-mode` `transcript.uid-remap`
`stable_contracts`	`env.required` `api.token` `api.invite-agent` `api.stop-conversation`

Quickstart Recipe Profile

This repo is a reusable quickstart sample for building browser voice-agent experiences with Agora Conversational AI Engine.

Recipe Role

Role: base quickstart recipe.
Target audience: developers bootstrapping a production-style Next.js voice agent app.
Reuse model: clone, bind project, run, then customize prompt/pipeline/UI.

Recipe Scope

This base recipe provides a copyable browser voice-agent starter with:

browser RTC audio and RTM event transport
server-side token, invite, stop, and optional custom LLM routes
default GeminiSTT (preview) → Gemini gemini-3.6-flash LLM → managed MiniMax TTS provider configuration using one Google key
pre-call, in-call, transcript, metrics, and connection-status UI

Baseline Implementation Guidance

This repository is the official Agora Next.js quickstart baseline for this recipe. Agents should use this repo's source and progressive disclosure docs as the starting point, then customize.

Do not recreate Agora ConvoAI integration from memory. Provider schemas, SDK builder fields, token behavior, and RTM event details can drift. For a new baseline implementation, follow L1/L2/from_scratch_bootstrap.md while copying verified patterns from this repo.

Extension Points

api.routes: add browser-facing routes under app/api, with shared request/response types in types/conversation.ts when the client consumes them.
prompts.system: edit ADA_PROMPT and GREETING in app/api/invite-agent/route.ts.
pipeline.providers: adjust the GeminiSTT, Gemini, and MiniMaxTTS builder chain. The defaults reuse NEXT_GOOGLE_API_KEY and use the standard AgoraClient.
ui.conversation: customize QuickstartPreCallCard, QuickstartConversationLayout, QuickstartTranscriptPanel, and QuickstartPipelineMetrics.

Invariants

Keep RtcTokenBuilder.buildTokenWithRtm for RTM-capable tokens.
Treat this repo as the official baseline; customize after preserving a working token, invite, RTC, RTM, and transcript flow.
Preserve StrictMode isReady guard for join/mic initialization.
Preserve UID remap (uid="0") and INTERRUPTED message-list inclusion.
Keep documentation synchronized when workflows/contracts change.

Stable Contracts

GET /api/generate-agora-token returns { token, uid, channel }.
POST /api/invite-agent accepts { requester_id, channel_name } and returns the agent id/state payload.
POST /api/stop-conversation accepts { agent_id }, stops the retained AgentSession, and treats a missing session as idempotent success.
Required env vars are NEXT_PUBLIC_AGORA_APP_ID, NEXT_AGORA_APP_CERTIFICATE, and NEXT_GOOGLE_API_KEY; the Google key serves GeminiSTT and Gemini LLM. Managed MiniMax TTS requires no additional credential.
components/LandingPage.tsx owns pre-call bootstrap and RTM client lifecycle.
components/ConversationComponent.tsx owns joined-session RTC/toolkit lifecycle.
lib/conversation.ts owns transcript normalization helpers.

Internal / Subject to Change

Visual styling and copy in the quickstart UI.
The exact reseller defaults for STT, LLM, and TTS models.
Connection issue display heuristics and metric chip presentation.

Consumer Onboarding Recipe

Clone or scaffold from template.
Bind Agora project and write .env.local.
Run pnpm run doctor and pnpm run dev.
Validate with pnpm run verify before sharing modifications.
Customize agent behavior and UI using the supported surfaces above.