← Back to Skills Marketplace
olina1ye

Ai Agent Builder Zh

by Olina1Ye · GitHub ↗ · v1.0.1 · MIT-0
cross-platform ✓ Security Clean
154
Downloads
0
Stars
1
Active Installs
2
Versions
Install in OpenClaw
/install ai-agent-builder-zh
Description
AI智能体开发 / AI Agent构建 / 智能体工作流 / agent builder。掌握工具调用、记忆管理、多步推理等核心能力,快速搭建客服机器人、数据分析代理、内容生成助手。适合工程师、产品经理、运营主管在企业自动化办公中应用,支持ChatGPT、Claude、文心一言等大模型集成。替代RPA重复劳动,...
README (SKILL.md)

AI 智能体构建指南

帮助工程师和产品经理设计、开发具备工具调用、记忆管理与多步骤推理能力的 AI 智能体(AI Agent),覆盖主流大模型(ChatGPT、Claude、文心一言等)集成模式,适配钉钉、企业微信、Web 等平台部署场景。

Tools Required

  • 代码编辑器(VS Code / Cursor)
  • LLM API(OpenAI / Anthropic / 百度千帆 / 阿里通义)
  • 工作流平台(n8n / Dify / FastGPT,可选)
  • 向量数据库(Milvus / Chroma,记忆模块可选)

Usage

  • "帮我设计一个能自动查询数据库并生成报表的 AI 智能体架构"
  • "我想用 Claude API 搭建一个客服机器人,支持多轮对话和工单创建,怎么设计工具调用流程?"
  • "如何给智能体加上长期记忆,让它记住用户的历史偏好和上下文?"

Examples

输入: 我需要一个销售助理智能体,能查询 CRM 系统里的客户信息、生成跟进邮件,并自动记录到飞书文档。请设计架构和工具列表。

输出:

智能体架构设计:销售助理 Agent

核心组件:

用户输入(销售需求)
      │
      ▼
  Agent(LLM 推理层:Claude / GPT-4)
      │
  ┌───┼───────────┐
  │   │           │
  ▼   ▼           ▼
查询  生成      记录工具
工具  工具      (飞书API)
(CRM) (邮件模板)

工具列表(Function Calling):

工具名 功能 输入参数
get_customer_info 查询 CRM 客户信息 customer_id / name
generate_followup_email 生成个性化跟进邮件 customer_info, context
save_to_feishu_doc 写入飞书文档 doc_id, content
search_deal_history 查询历史成交记录 customer_id, date_range

多步推理流程:

  1. 理解意图:解析销售人员输入,判断需要哪些工具
  2. 查询客户:调用 get_customer_info 获取基本信息 + search_deal_history 获取历史
  3. 生成邮件:基于客户画像调用 generate_followup_email,输出个性化内容
  4. 记录存档:调用 save_to_feishu_doc 将邮件草稿 + 操作记录写入飞书

记忆管理建议:

  • 短期记忆:对话上下文窗口(最近 10 轮)
  • 长期记忆:向量数据库存储客户偏好标签,每次对话前检索相关记忆注入 Prompt
  • 工作记忆:当前任务状态(已完成哪步、待执行哪步)存入 Redis

关键代码片段(Python + OpenAI Function Calling):

tools = [
    {
        "type": "function",
        "function": {
            "name": "get_customer_info",
            "description": "从CRM系统查询客户基本信息",
            "parameters": {
                "type": "object",
                "properties": {
                    "customer_name": {"type": "string", "description": "客户姓名或公司名"},
                },
                "required": ["customer_name"]
            }
        }
    },
    # 其他工具定义...
]

response = client.chat.completions.create(
    model="gpt-4o",
    messages=[{"role": "user", "content": user_input}],
    tools=tools,
    tool_choice="auto"  # 让模型自动决定调用哪个工具
)

部署建议:

  • 企业内部:通过企业微信 / 钉钉机器人接入,
Usage Guidance
This skill appears safe to install as an instruction-only guide. Before building systems from its examples, make sure any connected CRM, Feishu, LLM API, vector database, or workflow tool uses least-privilege access, avoids storing unnecessary personal or business data, and requires approval before high-impact writes or external actions.
Capability Analysis
Type: OpenClaw Skill Name: ai-agent-builder-zh Version: 1.0.1 The skill bundle is a purely educational guide for building AI agents, providing architectural advice and standard code snippets for OpenAI function calling. There are no executable scripts, network requests, or instructions that attempt to exfiltrate data or compromise the host environment (SKILL.md).
Capability Assessment
Purpose & Capability
The skill’s purpose is coherent: it teaches AI agent architecture, tool calling, memory, and LLM integration. Some examples involve business data access and document writes, which are sensitive if users later implement them.
Instruction Scope
The artifact provides design guidance and example code snippets rather than runnable automation. It mentions automatic tool selection, so implementations should add user approval and tool limits for high-impact actions.
Install Mechanism
There is no install specification, no required binaries, no required environment variables, and no code files; the static scan had nothing suspicious to analyze.
Credentials
References to LLM APIs, workflow platforms, and vector databases are expected for an AI agent builder guide, but users implementing the examples will need to manage credentials and data boundaries themselves.
Persistence & Privilege
The guide discusses long-term memory using vector databases and Redis; this is purpose-aligned but should be implemented with clear retention, consent, and deletion controls.
How to Use
  1. Make sure OpenClaw is installed (local or Docker)
  2. Run the install command in chat: /install ai-agent-builder-zh
  3. After installation, invoke the skill by name or use /ai-agent-builder-zh
  4. Provide required inputs per the skill's parameter spec and get structured output
Version History
v1.0.1
- 更新了描述与标签,更突出“agent builder”、“RPA 替代”、“企业AI解决方案”等关键词 - 增强了关键词覆盖和典型搜索场景,强调30%以上效率提升和自动化办公应用 - 用词更贴近企业实际需求,便于工程师、产品和运营主管快速定位并启用技能 - 其余内容保持一致,无核心技术变动
v1.0.0
Initial release: Comprehensive AI agent design and development guide in Chinese, covering core features and practical workplace applications. - Provides step-by-step instructions for building AI agents with tool calling, memory management, and multi-step reasoning. - Includes integration models for popular LLMs (ChatGPT, Claude, WenXin YiYan) and workflow platforms. - Sample workflows and architecture diagrams tailored to real-world Chinese enterprise scenarios (office automation, customer service bots, content assistants, data analysis). - Practical API and deployment recommendations for enterprise WeChat, DingTalk, Feishu, and web environments. - Example code snippets for tool integration and memory management.
Metadata
Slug ai-agent-builder-zh
Version 1.0.1
License MIT-0
All-time Installs 1
Active Installs 1
Total Versions 2
Frequently Asked Questions

What is Ai Agent Builder Zh?

AI智能体开发 / AI Agent构建 / 智能体工作流 / agent builder。掌握工具调用、记忆管理、多步推理等核心能力,快速搭建客服机器人、数据分析代理、内容生成助手。适合工程师、产品经理、运营主管在企业自动化办公中应用,支持ChatGPT、Claude、文心一言等大模型集成。替代RPA重复劳动,... It is an AI Agent Skill for Claude Code / OpenClaw, with 154 downloads so far.

How do I install Ai Agent Builder Zh?

Run "/install ai-agent-builder-zh" in the OpenClaw or Claude Code chat to install it in one step — no extra setup required.

Is Ai Agent Builder Zh free?

Yes, Ai Agent Builder Zh is completely free, licensed under MIT-0. You can download, install and use it at no cost.

Which platforms does Ai Agent Builder Zh support?

Ai Agent Builder Zh is cross-platform and runs anywhere OpenClaw / Claude Code is available (cross-platform).

Who created Ai Agent Builder Zh?

It is built and maintained by Olina1Ye (@olina1ye); the current version is v1.0.1.

💬 Comments