Web

Web is a CoreSpeed-hosted capability for grounding: it hands an agent current source material and leaves the reading to the agent. It needs no third-party account and runs behind the same /mcp seam as every other tool, so identity, organization holds, API-key caps, and activity recording apply unchanged.

Tools

Tool	Use it for
`web__search`	Ranked results with source URLs for a query, each with a short text snippet.
`web__scrape`	The readable text of one `https` page, freshly crawled.
`web__google_search`	Google's own results page for a query — organic ranking, People Also Ask, related searches, and optionally the AI Overview.
`web__google_maps_search`	Places on Google Maps — name, category, address, coordinates, phone, website, rating, review count, place ID.
`web__google_maps_reviews`	Reviews of one place — rating, text, date, likes, owner response.
`web__airbnb_listings`	Airbnb stays in a location — price, rating, guests, bedrooms, amenities, host, listing URL.
`web__tripadvisor_search`	TripAdvisor hotels, restaurants, and things to do in a place — rating, ranking, price level, address.
`web__tripadvisor_reviews`	Reviews of one TripAdvisor place — rating, text, travel date, trip type, owner response.
`web__amazon_products`	Amazon search results, or one product or category page — price, rating, reviews, availability, ASIN.
`web__zillow_listings`	Zillow homes in US ZIP codes, for sale, for rent, or sold — price, beds, baths, area, Zestimate.
`web__zillow_property`	One US home's full Zillow record by address or URL — price and tax history, Zestimate, facts, schools.

Search returns sources, not an answer

web__search does not synthesize. It returns ranked results — title, URL, published date, author, and by default a text snippet capped at 2,000 characters per result — and expects the agent to read them.

Parameter	Effect
`query`	What to search for. Required.
`num_results`	1–25, default 8.
`category`	Bias results toward one content type: `company`, `research paper`, `news`, `pdf`, `github`, `tweet`, `personal site`, `linkedin profile`, or `financial report`.
`include_text`	Inline a capped snippet per result. Default `true`.
`include_highlights`	Add query-relevant highlight sentences per result. Default `false`.
`livecrawl`	Force a fresh crawl of the results instead of the index cache — slower, fresher. Default `false`.

structuredContent

{
  "query": "MCP authorization spec RFC 9728",
  "resultCount": 2,
  "results": [
    {
      "title": "Authorization — Model Context Protocol",
      "url": "https://modelcontextprotocol.io/specification/…/authorization",
      "publishedDate": "2025-06-18",
      "author": null,
      "text": "…",
      "textTruncated": true
    }
  ]
}

textTruncated marks a snippet that hit the cap. To read a source in full, hand its URL to web__scrape.

Scrape reads one page

web__scrape fetches a single page and returns its clean text.

url must be https. IP literals, loopback, and internal hostnames are refused with invalid_input before any request leaves CoreSpeed.
max_characters defaults to 50,000 and is capped at 200,000. A clipped page sets textTruncated: true.
include_highlights adds query-agnostic highlight sentences from the page.

Every scrape prefers a live fetch: a cached copy is returned only while the site's own Cache-Control still marks it fresh, or for a few domains kept cached long-term. There is no multi-page crawl and no queue: for discovery across many pages, search first, then scrape the pages that matter.

Google, Maps, and listing sites

web__search finds sources; web__google_search answers a different question — what Google itself shows for a query, in what order. Use it for rankings, SERP features, and AI Overviews. The Maps, Airbnb, TripAdvisor, Amazon, and Zillow tools read those sites' own listings as structured records.

Tool	Parameters
`web__google_search`	`query` (Google operators work), `pages` 1–5 (~10 results each, default 1), `country` (ISO code), `language`, `site`, `time_range` (`hour` … `year`), `mobile`, `include_ai_overview`.
`web__google_maps_search`	`query` (what you would type into Maps), `location` (free text, city + country works best), `limit` 1–100 (default 20), `language`, `min_rating`, `skip_closed`, `include_details` (opening hours, popular times, review distribution).
`web__google_maps_reviews`	Exactly one of `place_url` or `place_id` (both in `web__google_maps_search` results), `limit` 1–100 (default 20), `sort` (`newest`, `most_relevant`, `highest`, `lowest`), `since` (with `sort` `newest`, the default), `keyword`, `language`. Reviewer identities are omitted.
`web__airbnb_listings`	`location`, `check_in` / `check_out` (YYYY-MM-DD), `adults`, `children`, `infants`, `pets`, `price_min` / `price_max`, `min_bedrooms`, `min_beds`, `min_bathrooms`, `currency`, `limit` 1–100 (default 20).
`web__tripadvisor_search`	`query` (a place, or a keyword), `kind` (`all`, `hotels`, `restaurants`, `attractions`), `language`, `currency`, `limit` 1–100 (default 20).
`web__tripadvisor_reviews`	`place_url` (from `web__tripadvisor_search`: `webUrl`), `rating`, `since`, `language`, `limit` 1–100 (default 20). Reviewer identities are omitted.
`web__amazon_products`	Exactly one of `query` (with `marketplace`, default `amazon.com`) or an Amazon `url` (product, search, or category), `limit` 1–100 (default 20).
`web__zillow_listings`	`zip_codes` (1–5 US ZIPs), `status` (`for_sale`, `for_rent`, `sold`), `price_min` / `price_max`, `days_on_zillow`, `limit` per ZIP code 1–40 (default 20; up to `limit` × ZIP codes in total).
`web__zillow_property`	Exactly one of `address` or a Zillow home `url` (from `web__zillow_listings`: `detailUrl`), plus `status` with a URL.

These tools gather their results live, which typically takes 10–60 seconds. They answer { "complete": true, "count": …, "results": [ … ] } and accept timeout_seconds (default 50, max 240) and max_characters. At the deadline the results gathered so far come back with "complete": false, and only those are billed; a collection that found nothing in time answers timeout, and one that fails upstream answers collection_failed — neither is charged.

Pricing and holds

Call	Charge
`web__search`	14 credits per request, covering up to 10 results with their snippets and highlights; 2 more per result above 10, counted on the `num_results` you ask for.
`web__scrape`	5 credits per page; 5 more with `include_highlights`.
`web__google_search`	6.75 credits per results page, plus 1.5 per run; 4.5 more per AI Overview captured.
`web__google_maps_search`	6 credits per place, plus a start fee of 0.075 per GB of run memory (about 0.3 per run); 1.5 more per place per filter (`min_rating`, `skip_closed`); 3 more per place with `include_details`.
`web__google_maps_reviews`	0.9 credits per review, plus the same start fee (about 0.075 per run).
`web__airbnb_listings`	6 credits per listing, plus 0.15 per run.
`web__tripadvisor_search`	7.5 credits per place, plus the start fee (about 0.15 per run).
`web__tripadvisor_reviews`	7.5 credits per review, plus the start fee (about 0.15 per run).
`web__amazon_products`	7.5 credits per product.
`web__zillow_listings`	3.6 credits per listing.
`web__zillow_property`	5.4 credits per property.

A failed request is never charged. The live figures are repeated in the tool descriptions your client shows from tools/list, and every charge lands itemized in the organization ledger — see Billing & credits. An organization billing hold or suspension stops a web call before it runs, as an isError result carrying payment_required or org_suspended; API-key spend caps apply the same way.

Controls

Web appears as Web search & scrape in Dashboard → Tools. Turning it off hides every web__* tool from tools/list for that member; see Capability controls.