Zero‑Code AI Agent with Higress: Build, Deploy, and Understand ReAct

What is an AI Agent

The rapid development of large‑model technology has revealed limitations such as insufficient domain knowledge and hallucinations, prompting the industry to propose the AI Agent concept (intelligent agent) that enables large models to think step‑by‑step, observe outcomes, and invoke tools to achieve a given goal.

What is an AI Gateway

An AI Gateway is an AI‑native API Gateway that extends traditional API‑gateway capabilities to meet AI requirements, such as token‑based rate limiting, multi‑model vendor support, A/B testing, and tracing.

Higress, an open‑source AI Gateway from Alibaba Cloud, builds on API Gateway and adds numerous AI‑related Wasm plugins to satisfy all AI‑native needs.

The author developed an AI Agent plugin on Higress, leveraging the gateway’s API management to enable zero‑code construction of an AI Agent that can call external services such as AMap (maps) and Seniverse (weather).

Using the AI Agent Plugin

API key acquisition – Register on the official sites of AMap and Seniverse, obtain free daily API keys, and reference the official documentation links.

Higress service configuration – In Higress route management, add DNS‑type services for the large‑model endpoint, AMap, and Seniverse.

Plugin Parameter Configuration

dashscope:
  # 通义千问大模型客户端配置
  apiKey: sk-xxxxxxxxxxxxxxxxxxxxxxx
  domain: dashscope.aliyuncs.com
  serviceName: dashscope
  servicePort: 443
promptTemplate:
  language: CH
apis:
  - apiProvider:
      domain: restapi.amap.com
      serviceName: geo
      servicePort: 80
      apiKey:
        in: query
        name: key
        value: fcxxxxxxxxxxxxxxxxxx
    api: |
      openapi: 3.1.0
      info:
        title: 高德地图
        description: 获取 POI 的相关信息
        version: v1.0.0
      servers:
        - url: https://restapi.amap.com
      paths:
        /v5/place/text:
          get:
            description: 根据POI名称，获得POI的经纬度坐标
            operationId: get_location_coordinate
            parameters:
              - name: keywords
                in: query
                description: POI名称，必须是中文
                required: true
                schema:
                  type: string
              - name: region
                in: query
                description: POI所在的区域名，必须是中文
                required: true
                schema:
                  type: string
        /v5/place/around:
          get:
            description: 搜索给定坐标附近的POI
            operationId: search_nearby_pois
            parameters:
              - name: keywords
                in: query
                description: 目标POI的关键字
                required: true
                schema:
                  type: string
              - name: location
                in: query
                description: 中心点的经度和纬度，用逗号隔开
                required: true
                schema:
                  type: string
  - apiProvider:
      domain: api.seniverse.com
      serviceName: seniverse
      servicePort: 80
      apiKey:
        in: query
        name: key
        value: SMxxxxxxxxxxxxxx
    api: |
      openapi: 3.1.0
      info:
        title: 心知天气
        description: 获取天气预报相关信息
        version: v1.0.0
      servers:
        - url: https://api.seniverse.com
      paths:
        /v3/weather/now.json:
          get:
            description: 获取指定城市的天气实况
            operationId: get_weather_now
            parameters:
              - name: location
                in: query
                description: 所查询的城市
                required: true
                schema:
                  type: string
              - name: language
                in: query
                description: 返回天气查询结果所使用的语言
                required: true
                schema:
                  type: string
                  default: zh-Hans
                  enum: [zh-Hans, en, ja]
              - name: unit
                in: query
                description: 表示温度的单位，有摄氏度和华氏度两种
                required: true
                schema:
                  type: string
                  default: c
                  enum: [c, f]

The configuration consists of three parts: dashscope (large‑model service), promptTemplate (custom ReAct template), and apis (external API definitions with OpenAPI specifications). The apis section is crucial for enabling the agent to call external tools.

Plugin Effects

Example curl requests to the gateway and the corresponding JSON responses illustrate map search and weather queries.

curl 'http://<gateway_ip>/api/openai/v1/chat/completions' \
  -H 'Accept: application/json, text/event-stream' \
  -H 'Content-Type: application/json' \
  --data-raw '{"messages":[{"role":"user","content":"我想在济南市鑫盛大厦附近喝咖啡，给我推荐几个"}],"model":"qwen","stream":false}'

{"content":" 在济南市鑫盛大厦附近，您可以选择以下咖啡店：
1. luckin coffee ..."}

curl 'http://<gateway_ip>/api/openai/v1/chat/completions' \
  -H 'Accept: application/json, text/event-stream' \
  -H 'Content-Type: application/json' \
  --data-raw '{"messages":[{"role":"user","content":"济南市现在的天气情况如何？"}],"model":"qwen","stream":false}'

{"content":" 济南市现在的天气状况为阴天，温度为31℃。"}

ReAct Principle

ReAct (Reasoning and Acting) combines chain‑of‑thought reasoning with tool‑calling actions and observations, as described in the paper “ReAct: Synergizing Reasoning and Acting in Language Models”. The model iteratively produces Thought, Action, Action Input, and Observation until a Final Answer is reached.

Plugin Implementation Logic

During the onHttpRequestBody stage, the original user query is replaced by a prompt that includes the tool definitions ({tools}, {tools_name}) and the custom ReAct template. The modified request is sent to the large‑model via ai‑proxy.

During the onHttpResponseBody stage, the model’s output is captured, stored in messageStore, and parsed with regular expressions to extract Action and Action Input. The corresponding external API is invoked (e.g., AMap location lookup), the result is appended as an Observation, and the updated conversation is sent back to the model. This loop repeats until the model emits a Final Answer, at which point proxywasm.ReplaceHttpResponseBody replaces the response body with the final answer.

Summary

The article introduced the background and concept of AI Agents, demonstrated how to use the AI Agent gateway plugin with Higress, showed its practical effects through example requests, and detailed the ReAct‑based implementation that enables zero‑code AI Agent construction.