Google search, news, maps, webpages, cited answers, research and ChatGPT/Gemini answers for agents.
Scored across 23 tools
The SERP verticals (search, images, news, videos, scholar, shopping, patents, maps/place) are cleanly separated, but six tools — ai_mode, answer, brave_ai, chatgpt, copilot, gemini — all do essentially the same thing (ask an engine, get a cited answer) differentiated only by provider, so an agent must guess which one to pick. crawl vs map vs extract vs webpage also sit close together, though their descriptions do draw usable lines (one page vs 20 URLs vs whole site vs URL discovery).
All names are lowercase snake_case, which is consistent in style, but the set mixes bare nouns (images, news, videos, patents) with verbs (search, extract, crawl) and opaque single words (answer, place). The map/maps pair is a real hazard: near-identical names for unrelated purposes (URL listing vs places).
23 tools is on the heavy end for what is fundamentally a search API, and the count is inflated by six interchangeable AI-answer endpoints that could be a single provider-parameterized tool. The async pairs (crawl/crawl_result, research/research_result) are justified, but overall the surface feels broad rather than tight.
The lifecycle is well covered: discovery (search/map/autocomplete), extraction (webpage/extract), async jobs with result-fetching companions (crawl, research), and all the major verticals. Minor gaps remain — no way to check credit balance or usage, and no cancel/abort for running crawl or research tasks, leaving users stuck on runaway jobs.
Ask Google AI Mode and get its answer with cited sources and products. 2 credits; takes 5-45 s; free if it fails.
Inputs: prompt (required), track, country, language, +1 more.
Answer a question from live Google results with numbered citations, written by Claude. depth=fast answers from the result page (3 credits); standard (default) also reads the top pages (8 credits at 3 sources). A failed answer is free.
Inputs: q (required), time, depth, country, +4 more.
Query suggestions. 0.5 credit.
Inputs: q (required), country, language.
Ask Brave Search's AI answer and get it with cited sources, the searches it ran and products. 2 credits; takes 1-10 s; free if it fails.
Inputs: prompt (required), track, country, language, +1 more.
Ask ChatGPT (chatgpt.com, as a logged-out user, searching the web) and get its answer as Markdown with the sources it cited, products and places it showed. For AI visibility: what ChatGPT tells people. 4 credits; takes 10-30 s; free if it fails. If it fails with upstream_timeout saying the answer is being finished in the background, call again with the same arguments within 10 minutes to receive it.
Inputs: prompt (required), track, country, web_search.
Ask Microsoft Copilot Search on Bing and get its answer with cited sources and products. 3 credits; takes 15-40 s; free if it fails.
Inputs: prompt (required), track, country, language.
Start an async crawl of one website: pages read as Markdown, following links up to max_depth hops (and the site's sitemaps), honouring robots.txt. With query, the most relevant pages are read first. Returns a task id; call crawl_result with it. 1 credit per page read; limit credits are reserved and the rest refunded.
Inputs: url (required), limit, query, max_depth, +3 more.
Get the status or the pages of a crawl started with crawl. Free.
Inputs: id (required).
Read up to 20 URLs (HTML or PDF) as clean Markdown. With query + highlights, return only the passages most relevant to the query; with summary, a short summary; with output_schema, structured JSON. 1 credit per URL read (+4 summary, +4 output_schema); failed URLs and failed AI steps are free.
Inputs: urls (required), query, prompt, country, +5 more.
Ask Google Gemini (gemini.google.com, as a logged-out user) and get its answer with the sources it grounded on. 3 credits; takes 5-30 s, a few up to 90; free if it fails. If it fails with upstream_timeout saying the answer is being finished in the background, call again with the same arguments within 10 minutes to receive it.
Inputs: prompt (required), track, country, language.
Search Google Images: image URL, size, title and source page. A page holds up to 100 images; page 2 starts at 101. 1 credit per page.
Inputs: q (required), num, page, engine, +7 more.
List a website's URLs (sitemaps + the start page's links), optionally ranked by a search phrase. 1 credit; free when nothing is found.
Inputs: url (required), limit, search, exclude_paths, +1 more.
Find places (name, rating, category, address, phone, website, place_id and cid, or oid with engine=yandex, and coordinates); the place tool returns one place's hours and full listing. Put the area in location (e.g. "Berlin, Germany" or "Kreuzberg, Berlin") or in q; ll searches around an exact point. 1 credit.
Inputs: q (required), ll, page, engine, +3 more.
Search Google News for recent articles. 1 credit.
Inputs: q (required), page, time, engine, +8 more.
Search patents. 1 credit.
Inputs: q (required), page, country, language.
The full listing of one place by its place_id (or cid/fid) from a maps result, or by oid from a Yandex Maps result: address, country, coordinates, time zone, categories, opening hours per weekday, website, phone, rating, description and photos. 1 credit; free when no place matches.
Inputs: cid, fid, oid, country, +2 more.
Start an async deep-research task: planned web searches, page reading and a cited Markdown report written by Claude. Returns a task id; call research_result with it. lite 30 / standard 75 / deep 200 credits, charged only when the report is written.
Inputs: input (required), effort.
Get the status or the report of a research task started with research. Free.
Inputs: id (required).
Search academic papers with citations and PDF links. 1 credit.
Inputs: q (required), page, country, language.
Search Google. Returns organic results, answer box, knowledge graph, People-also-ask and related searches as Markdown. 1 credit.
Inputs: q (required), num, page, time, +12 more.
Search products with prices and merchants. 1 credit.
Inputs: q (required), page, country, language, +1 more.
Recent tool additions, removals, and schema changes observed during successful MCP inspections.
Loading tool schema changelog…
Search Google straight from your AI agent. Web results, images, videos, news, products, scholarly ar
Web search, Google SERP, page extraction to markdown and cited answers for AI agents.
Real-time Google search (AI Overview, answer box, People also ask) and Google Maps places.
Google Search, News, Shopping, Images, Maps and AI Mode as data for AI agents.
Add one secure layer between your agents and this server.
