Search the web

You are wiring up the agntdata "Web" API to use this endpoint: Search the web.

1. CONNECT over MCP (preferred). Add the superagnt MCP server, then let the client
   run OAuth on first tool call — no token to paste:
     claude mcp add --scope user --transport http superagnt https://mcp.superagnt.com/mcp
     claude mcp login superagnt
   MCP endpoint: https://mcp.superagnt.com/mcp
   After connecting, the tool you want is named exactly:
     data_web_search
   Call agnt_tools_search or agnt_tools_list_enabled to confirm it is available.

2. REST fallback (if you are not using MCP):
     POST https://api.agntdata.dev/search
     Header: Authorization: Bearer <YOUR_AGNTDATA_API_KEY>
   Required parameters:
    - query (string, required) — The search query.
   Optional parameters:
    - limit (integer, optional) — Number of results per source. Default 10.
    - sources (array, optional) — Which result sources to query. Defaults to ['web']. Each entry is a preset string or an object with a `type`.
    - categories (array, optional) — Restrict results to these categories.
    - tbs (string, optional) — Time-based search filter (e.g. 'qdr:d' for past day).
    - location (string, optional) — Geo-target for the search (e.g. 'San Francisco, California, United States').
    - country (string, optional) — ISO country code. Default 'US'.
    - includeDomains (array, optional) — Only return results from these hostnames (mutually exclusive with excludeDomains).
    - excludeDomains (array, optional) — Exclude results from these hostnames (mutually exclusive with includeDomains).
    - timeout (integer, optional) — Request timeout in milliseconds. Default 60000.
    - scrapeOptions (object, optional) — When provided, each result is also scraped and its content returned. Accepts the same content options as /scrape (e.g. `formats`, `onlyMainContent`). Increases cost because every result is fetched.

3. COST & KEY. This endpoint is priced per call in credits, and each successful response reports its own cost in its meta.costCents field. Only successful calls are charged.
   Get an agntdata API key at https://app.agntdata.dev.

4. TEST. Connect the server (or set the key), discover the tool
   (data_web_search), run one minimal call with just the required parameters,
   and report back the shape of the JSON response (top-level fields).

Reference (machine-readable variant): https://superagnt.com/docs/apis/web/endpoints/search.md

{
  "name": "search",
  "description": "Runs a web search and returns result URLs + snippets — cheap; use this first and read the snippets. Add `scrapeOptions` ONLY when you actually need the full text of the pages, because it then downloads and bills EVERY result — keep `limit` small when you do (a few results, not dozens). To read specific pages, prefer searching first and then scraping just the one or two URLs you chose. Supports web, news, and image sources.",
  "parameters": {
    "type": "object",
    "properties": {
      "query": {
        "type": "string",
        "maxLength": 500,
        "description": "The search query."
      },
      "limit": {
        "type": "integer",
        "description": "Number of results per source. Default 10.",
        "minimum": 1,
        "maximum": 100,
        "default": 10
      },
      "sources": {
        "type": "array",
        "description": "Which result sources to query. Defaults to ['web']. Each entry is a preset string or an object with a `type`.",
        "items": {
          "oneOf": [
            {
              "type": "string",
              "enum": [
                "web",
                "news",
                "images"
              ]
            },
            {
              "type": "object",
              "required": [
                "type"
              ],
              "properties": {
                "type": {
                  "type": "string",
                  "enum": [
                    "web",
                    "news",
                    "images"
                  ]
                }
              }
            }
          ]
        }
      },
      "categories": {
        "type": "array",
        "items": {
          "type": "string",
          "enum": [
            "github",
            "research",
            "pdf"
          ]
        },
        "description": "Restrict results to these categories."
      },
      "tbs": {
        "type": "string",
        "description": "Time-based search filter (e.g. 'qdr:d' for past day)."
      },
      "location": {
        "type": "string",
        "description": "Geo-target for the search (e.g. 'San Francisco, California, United States')."
      },
      "country": {
        "type": "string",
        "description": "ISO country code. Default 'US'.",
        "default": "US"
      },
      "includeDomains": {
        "type": "array",
        "items": {
          "type": "string"
        },
        "description": "Only return results from these hostnames (mutually exclusive with excludeDomains)."
      },
      "excludeDomains": {
        "type": "array",
        "items": {
          "type": "string"
        },
        "description": "Exclude results from these hostnames (mutually exclusive with includeDomains)."
      },
      "timeout": {
        "type": "integer",
        "description": "Request timeout in milliseconds. Default 60000.",
        "minimum": 1000
      },
      "scrapeOptions": {
        "type": "object",
        "description": "When provided, each result is also scraped and its content returned. Accepts the same content options as /scrape (e.g. `formats`, `onlyMainContent`). Increases cost because every result is fetched.",
        "properties": {
          "formats": {
            "type": "array",
            "items": {},
            "description": "Output formats for each scraped result (e.g. ['markdown'])."
          },
          "onlyMainContent": {
            "type": "boolean",
            "default": true
          }
        }
      }
    },
    "required": [
      "query"
    ]
  }
}