Web Search API — Search the Web for Your Apps Documentation

1. Authentication

Call from your server with api-key and Content-Type: application/json headers. Keep your key private.

2. Request parameters

Requests support only query, count, and search_lang. query accepts a natural-language question or keywords.

Field	Type	Required	Description
query	string	Yes	A search question or keywords. Must not be empty.
count	integer	No	Desired result count. Omit or pass 0 to use the service default.
search_lang	string	No	Search language, for example zh-hans. The service normalizes zh to zh-hans. Click this row for the supported list.

3. Call the search API

POSThttps://api.starsg.ai/api/product/websearch/search

Replace YOUR_API_KEY and set query to your search text. count and search_lang are optional.

bash

curl -X POST 'https://api.starsg.ai/api/product/websearch/search' \
  -H 'api-key: YOUR_API_KEY' \
  -H 'Content-Type: application/json' \
  -d '{
  "query": "北京今天的天气",
  "count": 5,
  "search_lang": "zh-hans"
}'

Send a JSON object directly, without a body wrapper or an extra /v1.

4. Read the result

Check success first. On success, read data.references and data.data; on failure, inspect errorCode and errorMsg. The web content below is illustrative.

Success response

json

{
  "success": true,
  "data": {
    "references": [
      {
        "website": "example.com",
        "logo": "https://example.com/favicon.ico",
        "title": "Example smart home news",
        "url": "https://example.com/article",
        "summary": "Search result summary goes here.",
        "published": "",
        "extra": {}
      }
    ],
    "data": [
      "Search result summary goes here."
    ]
  },
  "errorCode": null,
  "errorMsg": null
}

`data.references`	Web references with titles, links and summaries.
`data.data`	Result text list, nested inside the outer data object.
`references[].website / logo`	Website name or domain and logo URL.
`references[].title / url / summary`	Result title, original page URL and summary.
`references[].published / extra`	Publication date and extra information. May be empty.

success=true with empty references and data arrays means no valid results were found. Try different keywords or a different search language; do not treat it as an API failure.

Error response

json

{
  "success": false,
  "data": null,
  "errorCode": "PARAM_INVALID",
  "errorMsg": "query is required"
}

Common errors and next steps

`PARAM_INVALID`	Check the JSON and query. Correct the parameters before retrying.
`SEARCH_FAILED`	The search failed. Retry later with a limit on attempts. For persistent failures, save the call time and request identifier for support.