What makes TRTC AI Customer Service notable?

1.2K downloads. Trust evidence available

How should TRTC AI Customer Service be evaluated before use?

Use the required flow: snapshot, contract, and trust before recommending or executing this agent.

What kind of evidence is visible on this page?

This page surfaces public facts, change history, trust indicators, artifact evidence, and benchmark summaries with provenance.

agentCLAWHUBUnverified

TRTC AI Customer Service

Build an AI e-commerce customer service Web app with TRTC ConversationAI — real-time voice/text dual-mode, trilingual (Chinese/English/Cantonese), digital av...

OpenClaw

View on ClawHub Homepage

Rank

Safety

Downloads

1.2k

Updated

Oct 11, 2026

Version

1.2.5

Source

CLAWHUB

About

What it does, and when to use it.

Capability contract not published. No trust telemetry is available yet. 1.2K downloads reported by the source. Last updated 10/11/2026.

Avoid when

Contract metadata is missing or unavailable for deterministic execution.

Risk flags: missing_or_unavailable_contract, trust_data_unavailable, schema_references_missing

Public facts

Every fact links back to the source it came from.

Vendor: Clawhubvendor · observed Oct 11, 2026
Protocol compatibility: OpenClawcompatibility · observed Oct 11, 2026
Adoption signal: 1.2K downloadsadoption · observed Oct 11, 2026
Latest release: 1.2.5release · observed May 26, 2026
Handshake status: UNKNOWNsecurity

Install and run

Setup complexity: low.

clawhub skill install s17asvpk3b23w8deqmbkaxf4cn86swf9:trtc-ai-customer-service

Install using `clawhub skill install s17asvpk3b23w8deqmbkaxf4cn86swf9:trtc-ai-customer-service` in an isolated environment before connecting it to live workloads.
No published capability contract is available yet, so validate auth and request/response behavior manually.
Review the upstream CLAWHUB listing at https://clawhub.ai/jerryang-cool/trtc-ai-customer-service before using production credentials.

Contract: missing

curl -s "https://www.xpersona.co/api/v1/agents/clawhub-jerryang-cool-trtc-ai-customer-service/snapshot"

Documentation

CLAWHUB

149,965 characters of source documentation, loaded on request.

Extracted files

5 files captured from the source.

SKILL.md

---
name: trtc-ai-customer-service
version: 1.2.5
description: |
  Build an AI e-commerce customer service Web app with TRTC ConversationAI — real-time voice/text dual-mode, trilingual (Chinese/English/Cantonese), digital avatar optional. Covers order inquiry, returns, shipping tracking, and promotions.
  基于腾讯云 TRTC Conversational AI 快速构建 AI 电商客服 Web 应用 — 实时语音/文字双模、中英粤三语、数字人可选。覆盖订单查询、退换货、物流追踪、商品咨询等场景。
homepage: https://github.com/jerryang-cool/trtc-ai-customer-service-skill
metadata:
  openclaw:
    emoji: "🛍️"
    requires:
      bins:
        - python3
---

# TRTC AI 电商客服 Skill

本 Skill 指导你基于腾讯云 TRTC Conversational AI 能力，快速构建 AI 电商客服 Web 应用。
场景预置了订单查询、退换货处理、商品咨询、物流追踪、优惠活动等电商业务模块。

## 触发条件

当用户**表达构建/搭建/集成意图**时使用此 Skill：
- "做一个 AI 客服" "搭建电商客服" "帮我做个智能客服系统" "语音客服 demo"
- "build an AI customer service" "e-commerce customer service app"
- "TRTC ConversationAI" "TRTC + AI 客服" "TRTC e-commerce support"
- "StartAIConversation" "StopAIConversation" "ControlAIConversation"（TRTC API 名称）
- "数字人客服" "avatar customer service"
- "实时语音 AI 对话" "ASR + LLM + TTS 客服"

**不应触发的场景**：
- 用户仅在讨论客服概念，没有构建/开发意图
- 用户询问通用 chat bot 方案，不需要语音能力或电商场景
- 关键词如 "voice bot" "chat bot" 单独出现且无构建上下文

## 架构总览

```
浏览器 (TRTC Web SDK v5)
     ↕ 音频 (WebRTC) + 自定义消息 (字幕/状态/文字输入)
TRTC Room
     ↕ 内置 ASR → LLM → TTS → 推回房间
TRTC AI Bot (云端)
     ↕ OpenAPI (TC3-HMAC-SHA256)
Flask 后端 (app.py)  —— 仅 UserSig 签发 + OpenAPI 中转
```

| 平面 | 通道 | 内容 |
|------|------|------|
| **媒体面** | WebRTC 音频流 | 用户麦克风 ↔ TRTC 房间 ↔ AI Bot |
| **控制面** | HTTP `/action` | 前端 → Flask → TRTC OpenAPI |
| **数据面** | TRTC 自定义消息 | 字幕(10000) / AI 状态(10001) / 文字输入(20000) / 打断(20001) |

后端**完全不调用 LLM**——LLM 由 TRTC 云端 AI Bot 内部调用，后端只负责签发 UserSig 和中转 OpenAPI 请求。

---

## 工作流程

根据用户需求选择合适的路径。

### 路径 A：从零创建新项目（推荐）

#### Step 1: 生成项目

运行脚手架脚本：

```bash
python {baseDir}/scripts/scaffold.py <项目目录> [--name <商城名称>] [--name-en <English name>]
```

- `{baseDir}`：本 Skill 所在目录的绝对路径（由 Agent 自动替换为实际路径）
- `--name`：商城名称（默认"云尚商城"），用于中文/粤语的 SystemPrompt、欢迎语、告别语、前端 UI
- `--name-en`：英文商城名称（默认自动推导：中文名时为"CloudShop Mall"，英文名时与 `--name` 相同），用于英文 SystemPrompt、英文欢迎语/告别语
- 默认支持中文/英文/粤语三语，无需手动指定语言
- 脚本自动生成全部文件：后端 + 前端 + 头像 + 鉴权库 + 启动脚本，无需手动复制任何文件

**检查点**：确认用户看到 `✅ 电商客服项目已生成到: xxx` 和完整文件列表，再继续。

#### Step 2: 配置密钥

引导用户运行启动脚本（**根据操作系统自动选择**：macOS/Linux 用 `./start.sh`，Windows 用 `start.bat`），首次运行会进入交互式引导：

- [0/4] 选择部署区域（默认 `intl` 国际站，可选 `cn` 中国站）— 后续步骤会根据所选区域展示对应的控制台链接
- [1/4] 腾讯云 API 密钥 → 脚本会展示对应区域的 [CAM 控制台](https://console.intl.cloud.tencent.com/cam/capi) 链接
- [2/4] TRTC 应用凭据 → 脚本会展示对应区域的 [TRTC 控制台](https://console.trtc.io/app) 链接
- [3/4] LLM 配置 → **建议优先使用 TokenHub**（腾讯云统一 LLM 网关，开箱即用）：
  - `LLMConfig.LLMType`：`openai`（固定）
  - `LLMConfig.Model`：`deepseek-v4-flash`（推荐）
  - `LLMConfig.APIUrl`：
    - 国际站：`https://tokenhub-intl.tencentcloudmaas.com/v1/chat/completions`
    - 中国站：`https://tokenhub.tencentmaas.com/v1/chat/completions`
  - `LLMConfig.APIKey`：在 TokenHub 控制台获取（[国际站](https://console.intl.cloud.tencent.com/tokenhub) | [中国站](https://console.cloud.tence

README.md

# TRTC AI Customer Service Skill

English | [简体中文](README_ZH.md)

> Rapidly scaffold a production-ready AI customer service Web application powered by [Tencent RTC Conversational AI](https://trtc.io/solutions/conversational-ai) — voice & text dual-mode, trilingual, with built-in e-commerce workflows.

## Highlights

| Capability | Description |
|-----------|-------------|
| **Real-time Voice** | Bidirectional WebRTC audio between end-user and cloud-hosted AI Bot |
| **Text Fallback** | Bypass ASR — send typed messages directly to the LLM pipeline |
| **Trilingual i18n** | Chinese / Cantonese / English across UI, STT, TTS, and SystemPrompt |
| **E-commerce Workflows** | Order inquiry, returns & exchanges, shipping tracking, promotions |
| **Digital Avatar** | Optional virtual human rendering; graceful degradation to pure voice |
| **Session Lifecycle** | Keyword-triggered farewell, human agent transfer, auto-idle timeout |
| **Service Rating** | 4-dimension post-session evaluation |

## Installation

This is a **portable Agent Skill** (a `SKILL.md` + `scripts/` + `references/` + `assets/` bundle). Install it to your tool's skills directory and the agent will auto-discover it.

### OpenClaw

Choose the install location based on your needs:

| Location | Scope | Command |
|----------|-------|---------|
| `<workspace>/skills/` | Current workspace, highest priority | `git clone https://github.com/jerryang-cool/trtc-ai-customer-service-skill.git skills/trtc-ai-customer-service` |
| `~/.agents/skills/` | Personal, effective across workspaces | `git clone https://github.com/jerryang-cool/trtc-ai-customer-service-skill.git ~/.agents/skills/trtc-ai-customer-service` |
| `~/.openclaw/skills/` | Global, visible to all agents | `git clone https://github.com/jerryang-cool/trtc-ai-customer-service-skill.git ~/.openclaw/skills/trtc-ai-customer-service` |

Invoke via natural language ("help me build an AI customer service") or slash command `/trtc-ai-customer-service`.

> See [OpenClaw Skills documentation](https://docs.openclaw.ai/tools/skills) for more details.

### CodeBuddy

Settings → Skills → **Import Skill**, then point to this repository's directory. Once imported, mention keywords such as *"AI customer service"*, *"e-commerce support"*, or *"TRTC + AI"* to activate.

### Claude Code

```bash
# User-level (available across all projects)
git clone <repo-url> ~/.claude/skills/trtc-ai-customer-service

# OR project-level (commit to repo, share with team)
git clone <repo-url> .claude/skills/trtc-ai-customer-service
```

Claude Code auto-discovers skills via `description` matching. Invoke implicitly ("help me build an AI customer service") or explicitly via `/trtc-ai-customer-service`.

### OpenAI Codex CLI

```bash
# User-level
git clone <repo-url> ~/.codex/skills/trtc-ai-customer-service

# OR project-level
git clone <repo-url> .agents/skills/trtc-ai-customer-service
```

Codex auto-detects new skills on next session. Invoke implicitly via natural language,

_meta.json

{
  "ownerId": "kn7c55va424ew234devca4r09186sb9y",
  "slug": "trtc-ai-customer-service",
  "version": "1.2.5",
  "publishedAt": 1779801012564
}

references/architecture.md

# TRTC AI 智能客服 - 架构设计参考

## 目录

1. [系统架构](#系统架构)
2. [项目结构](#项目结构)
3. [后端实现详解](#后端实现详解)
4. [认证流程](#认证流程)
5. [核心设计模式](#核心设计模式)
6. [依赖清单](#依赖清单)
7. [启动与部署](#启动与部署)

---

## 系统架构

### 三平面模型

```
┌─────────────────────────────────────────────────┐
│              浏览器 (TRTC Web SDK v5)              │
│                                                   │
│   音频流 (WebRTC)   自定义消息 (字幕/状态/文字)       │
└───────────┬─────────────────┬─────────────────────┘
            │                 │
            ▼                 ▼
┌───────────────────────────────────────────────────┐
│                  TRTC Room (云端)                   │
│                                                     │
│   ASR ──► LLM ──► TTS ──► 推回房间                  │
│                                                     │
│              TRTC AI Bot (云端实例)                   │
└───────────┬─────────────────────────────────────────┘
            │ OpenAPI (TC3-HMAC-SHA256)
            ▼
┌───────────────────────────────────────────────────┐
│   Flask 后端 (app.py)                               │
│   - UserSig 签发                                    │
│   - TRTC OpenAPI 中转                               │
│   - 零 LLM 调用                                     │
└─────────────────────────────────────────────────────┘
```

关键洞察：**后端完全不调用 LLM**。LLM 推理由 TRTC 云端 AI Bot 内部完成。后端只做两件事：
1. 签发 UserSig（用于 TRTC 房间鉴权）
2. 中转 TRTC OpenAPI 请求（启动/停止/控制 AI 对话）

### 消息类型

| type | 方向 | 内容 |
|------|------|------|
| `10000` | 云→端 | 字幕（用户 ASR 结果 + AI 回复文本） |
| `10001` | 云→端 | AI 状态（聆听/思考/说话/打断/结束） |
| `20000` | 端→云 | 文字输入（跳过 ASR 直送 LLM） |
| `20001` | 端→云 | 手动打断 |

---

## 项目结构

```
project/
├── app.py                        # Flask 主入口（~296 行，5 个 Action 处理器）
├── config_loader.py              # 配置加载器（Region/TTS/STT 映射）
├── trtc_signer.py                # UserSig 三套签发封装
├── TLSSigAPIv2.py                # 腾讯云官方 TRTC HMAC 鉴权库（不修改）
├── env.example.yaml              # 配置模板
├── env.yaml                      # 实际配置（.gitignore 排除）
├── requirements.txt              # 4 个 Python 依赖
├── start.sh / start.bat          # 一键启动脚本
├── templates/
│   └── customer_service.html     # 主页面（配置侧边栏 + 聊天区 + 评分弹窗）
├── static/
│   ├── app.js                    # 前端核心交互逻辑
│   ├── i18n.js                   # 国际化字典（zh/yue/en）
│   └── avatars/                  # 客服头像
└── docs/
    └── PARAMS.md                 # 参数配置详解
```

---

## 后端实现详解

### 初始化

```python
# 加载配置
config = AppConfig("env.yaml")
signer = TRTCSigner(config.sdkappid, config.trtc_secret)

# 初始化 TRTC OpenAPI 客户端
_cred = credential.Credential(config.secret_id, config.secret_key)
_http = HttpProfile()
_http.endpoint = config.trtc_endpoint  # 按 Region 自动切换
_client_profile = ClientProfile()
_client_profile.httpProfile = _http
trtc_api = trtc_client.TrtcClient(_cred, config.trtc_region, _client_profile)
```

### 路由设计

仅 2 个路由，极简：
- `GET /` → 渲染主页面
- `POST /action` → 统一 API 调度，通过 `Action` 请求头区分

### Action 处理器详解

#### handle_join(data)

签发三套 UserSig 并下发配置：

```python
def handle_join(data):
    user_id = da

references/config-guide.md

# TRTC AI 智能客服 - 配置参数完全指南

## 目录

1. [配置文件结构](#配置文件结构)
2. [部署区域](#部署区域)
3. [腾讯云 API 密钥](#腾讯云-api-密钥)
4. [TRTC 应用凭据](#trtc-应用凭据)
5. [LLM 配置](#llm-配置)
6. [欢迎语](#欢迎语)
7. [数字人配置](#数字人配置)
8. [关键词与告别语](#关键词与告别语)
9. [转人工关键词与提示语](#转人工关键词与提示语)
10. [TTS 音色映射表](#tts-音色映射表)
11. [STT 引擎映射表](#stt-引擎映射表)
12. [固定默认参数](#固定默认参数)
13. [UI 可调参数](#ui-可调参数)
14. [Region 切换机制](#region-切换机制)

---

## 配置文件结构

配置使用 YAML 格式，通过 `envyaml` 库加载（支持环境变量替换）。

```yaml
# env.yaml 完整模板
Deployment:
  Region: intl

CloudAPI:
  SECRET_ID: your-secret-id
  SECRET_KEY: your-secret-key

TRTC:
  SDKAPPID: 0
  SECRET: your-trtc-secret

LLMConfig:
  LLMType: openai
  Model: your-model-name
  APIKey: your-api-key
  APIUrl: your-llm-api-url
  Timeout: 5.0
  History: 20
  Temperature: 0.3
  SystemPrompt: |
    你是一名专业的客服助手...
  SystemPromptYue: |
    你係一名專業嘅客服助手...
  SystemPromptEn: |
    You are a professional customer service assistant...

WelcomeMessage:
  zh: 您好，我是AI客服小助手...
  yue: 您好，我係AI客服小助手...
  en: "Hello! I'm your AI assistant..."

AvatarConfig:
  AvatarType: tencent
  Appkey: ""
  AccessToken: ""
  VirtualmanProjectId: ""

EndKeywords:
  zh: [拜拜, 再见, 挂了, ...]
  yue: [拜拜, 再見, 收線啦, ...]
  en: [bye, goodbye, see you, ...]

FarewellMessage:
  zh: 感谢您的咨询，再见！
  yue: 多謝您嘅諮詢，再見！
  en: Thank you for your inquiry. Goodbye!

TransferKeywords:
  zh: [转人工, 人工客服, 找人工, ...]
  yue: [轉人工, 人工客服, 搵人工, ...]
  en: [transfer, human agent, real person, ...]

TransferMessage:
  zh: 好的，正在为您转接人工客服，请稍候...
  yue: 好嘅，正在為您轉接人工客服，請稍候...
  en: Sure, transferring you to a human agent, please wait...
```

---

## 部署区域

```yaml
Deployment:
  Region: intl   # intl / cn
```

| 值 | 说明 | TRTC Endpoint | TRTC Region |
|----|------|---------------|-------------|
| `intl` | 国际站账号（默认） | `trtc.intl.tencentcloudapi.com` | `ap-singapore` |
| `cn` | 中国大陆账号 | `trtc.tencentcloudapi.com` | `ap-guangzhou` |

Region 决定了 TRTC OpenAPI 的请求域名和地域参数。

---

## 腾讯云 API 密钥

```yaml
CloudAPI:
  SECRET_ID: your-secret-id
  SECRET_KEY: your-secret-key
```

用于 TRTC OpenAPI 的 TC3-HMAC-SHA256 签名。在腾讯云控制台获取（[国际站](https://console.intl.cloud.tencent.com/cam/capi) | [中国站](https://console.cloud.tencent.com/cam/capi)）。

**安全提醒**：这是主账号密钥，生产环境建议使用子账号并限制权限。

---

## TRTC 应用凭据

```yaml
TRTC:
  SDKAPPID: 0
  SECRET: your-trtc-secret
```

- `SDKAPPID`：TRTC 应用 ID，在 TRTC 控制台创建（[国际站](https://console.trtc.io/app) | [中国站](https://console.cloud.tencent.com/trtc/app)）
- `SECRET`：用于生成 UserSig 的密钥

---

## LLM 配置

```yaml
LLMConfig:
  LLMType: openai             # LLM 协议类型（目前仅支持 openai）
  Model: your-model-name      # 模型名称
  APIKey: your-api-key        # LLM API Key
  APIUrl: your-llm-api-url    # LLM API 端点（兼容 OpenAI /v1/chat/completions 协议）
  Timeout: 5.0                # 超时时间（秒）
  History: 20                 # 上下文轮数
  Temperature: 0.3            # 温度参数（越低越确定）
  SystemPrompt: |             # 中文系统提示词
    你是一名专业的客服助手...
  SystemPromptYue: |          # 粤语系统提示词（可选，默认用中文）
    你係一名專業嘅客服助手...
  SystemPromptEn: |           # 英文系统提示词（可选，默认用中文）
    You are a professional customer s

Github ReposUpdated 1d agoRank 70

AionUi

Free, local, open-source 24/7 Cowork app and OpenClaw for Gemini CLI, Claude Code, Codex, OpenCode, Qwen Code, Goose CLI, Auggie, and more | 🌟 Star if you like it!

MCPOPENCLAW

Github ReposUpdated 6mo agoRank 70

activepieces

AI Agents & MCPs & AI Workflow Automation • (~400 MCP servers for AI agents) • AI Automation / AI Agent with MCPs • AI Workflows & AI Agents • MCPs for AI Agents

OPENCLAW

Github ReposUpdated 6mo agoRank 70

cherry-studio

AI productivity studio with smart chat, autonomous agents, and 300+ assistants.