Skip to main content
Glama
650,501 tools. Updated 2026-10-14 00:38

"Using Google Search for Answer Generation" matching MCP tools:

  • Google Images search for AI agents at $0.010 per call, from the same Serper.dev source as /web/search. Send a query, get back compact JSON: top image results with position, title, image URL, source page link and dimensions. Tune with num (1-10 results), country and language (2-letter codes). Zero results is a valid, honest answer. Pay per call in USDC on Base, no account, no API key.
    ConnectorNo auth
  • List supported Google Maps place type values for search filters. Returns place_types as a string array. Use a value with place_type on google-maps.search or google-maps.nearby_search. Cost = 1 token.
    ConnectorNo auth
  • List Search Console properties for this account. Returns Vee3-managed domains plus sites added with google-search-console.add_site. Call this to discover sites, then omit site_url on later calls to use the default property, or pass an exact site_url from this response. If no properties appear, add a site with google-search-console.add_site or register a domain with domains.register. Cost = 5 tokens.
    ConnectorNo auth
  • Discover the AdKit MCP contract with progressive drill-down. Top-level paths: "manage" (campaigns, accounts, and reporting across Meta, Google, TikTok, Reddit, X, and LinkedIn, plus Microsoft and ChatGPT Ads raw access and drafts), "integrations" (read-only GA4 data), "library" (browse ads, advertisers, and inspiration boards), and "studio" (AI static ad generation). Start with adkit_help() for the root overview, or drill down like adkit_help({ path: "manage google campaigns create" }). A keyword search with q, e.g. adkit_help({ q: "keyword planner" }) → "manage google research keywords", covers the full command catalog: a capability missing from one drill-down path may still exist under another.
    ConnectorAPI key
  • Save the gold reference for an enrichment or schema-generation benchmark. Only set reference_verified=true after checking its values against trusted evidence or obtaining human sign-off; a generated answer alone is not verification. Schema-generation references must be GeneratedJsonSchema objects. Sample-generation scenarios reject references because they are rubric-scored. Requires owner and a plan with benchmarks; no LLM call. A verified reference enables run_benchmark. See enricher://docs/model-benchmark.
    ConnectorOAuth
  • Ask Google AI Mode a prompt and get its structured answer (reply_parts: paragraphs, headings, lists, images) with reference_links citations. Pass session_token from a previous response to ask a follow-up in the same conversation. Answers can take over a minute. Use search_web for plain search results. Billed $0.005 per request.
    ConnectorNo auth

Matching MCP Servers

Matching MCP Connectors

  • google search: google web search api, web, images, videos, news, music, favicon, proxy, audio.

  • Scrape Google search results with SERP data, ads, and knowledge panels

  • List a model provider's voices so you can pass the right id to a generation tool. Default 'elevenlabs-voices' (the voices for generate_audio); also covers Kling and Google Veo voices. Returns each voice's id and name. (This is Unifically's resource catalog; it is not the MCP protocol's resources/list.)
    ConnectorNo auth
  • List the model catalog (video, image, audio, upscale) with the short aliases you can pass as 'model'. Filter by category and/or provider (e.g. provider: 'google'). Use the returned ids or aliases on the generation tools. Refreshed hourly from the live catalog.
    ConnectorNo auth
  • Draft adding or removing services, using ids from my_service_options, plus custom services Google does not list. Every service not named stays on the listing. Creates a DRAFT only — nothing on Google changes. Show the owner the draft this returns, word for word, and publish it with publish_change only after they say yes.
    ConnectorNo auth
  • Fallback people search through Google, e.g. when search_linkedin_people is protected or times out. Not for general web research (use search_web). Google X-Ray search for public LinkedIn profiles via Google operators. For company discovery pass the company in keywords and several realistic jobTitles; they are combined with OR in one web query. Existing matching contacts are returned with their contact_id instead of disappearing as duplicates, so they can be enriched. Results remain unverified until enrich_contacts confirms the current employer.
    ConnectorAPI key
  • Get details for one sitemap submitted to Search Console. Call google-search-console.list_sites to discover sites, then google-search-console.list_sitemaps for an exact feed_path when available. Omit site_url to use the default property. Returns submission status, download time, warnings, errors, and content counts. Cost = 5 tokens.
    ConnectorNo auth
  • Audits content for Answer Engine Optimization (AEO): Evaluates concise 40-60 word direct answer definition blocks, question-based H2/H3 subheadings (What/How/Why), FAQ schema alignment, and citation readiness for Google AI Overviews and Perplexity. USAGE GUIDELINES: - Use when optimizing content to win conversational AI search citations, direct answers, and Perplexity summaries. - Do NOT use for brand knowledge-graph entity reconciliation; use 'seo_audit_geo' instead. - Do NOT use for standard on-page metadata; use 'seo_audit_onpage' instead. BEHAVIORAL TRANSPARENCY: - Safe, read-only diagnostic evaluation. No file modifications.
    ConnectorNo auth
  • Google web search via Serper: returns organic results (title, link, snippet, position), an answer box if present, and a knowledge graph summary. Accepts query, result count (max 20), and optional country code (gl) and location for geo-targeted results.
    ConnectorNo auth
  • Exact Google Ads search volume for `<keyword>` — Google's own monthly search-volume numbers (plus competition and CPC) from the Ads API, for up to 10 keywords. Use when you specifically need Google Ads figures; for general SEO volume + keyword difficulty, prefer seo_keyword_overview (cheaper). Example: seo_keyword_google_ads_volume({ keywords: ["running shoes"], location_code: 2840, _apiKey: "your-base64-key" })
    ConnectorNo auth
  • Search the web using ScrapingBee's Fast Search API and return structured results. This is the PRIMARY tool for web searches — it returns clean organic results and top stories in under one second at a lower credit cost. USE THIS TOOL FIRST for any general web search. Only fall back to get_google_search_results if you need specialized search types such as news, maps, Google Lens, shopping, image search, or Google AI mode. Scope: one query per call. To run many queries in one pass, or to write results to disk instead of into the conversation, use the ScrapingBee CLI — `scrapingbee <command> --input-file queries.txt --output-dir results`.
    ConnectorAPI key
  • Search the web using ScrapingBee's Fast Search API and return structured results. This is the PRIMARY tool for web searches — it returns clean organic results and top stories in under one second at a lower credit cost. USE THIS TOOL FIRST for any general web search. Only fall back to get_google_search_results if you need specialized search types such as news, maps, Google Lens, shopping, image search, or Google AI mode. Scope: one query per call. To run many queries in one pass, or to write results to disk instead of into the conversation, use the ScrapingBee CLI — `scrapingbee <command> --input-file queries.txt --output-dir results`.
    ConnectorNo auth
  • List all Google Search Console properties (sites) connected to this account. Requires Google Search Console to be connected. Direct the user to rankparse.com/dashboard/integrations to connect it. When several Google accounts are connected, the properties visible to any of them are listed once each.
    ConnectorNo auth
  • Use answer_query to get a grounded answer to a query about Google developer products. This tool has limited quota. This tool will synthesize information from the corpus to generate an answer to the query. answer_query grounds answers using the same corpus as search_documents. This tool returns the generated answer_text and a list of document names (references) used to generate the answer. Use get_documents with the document names to fetch the entire document content if needed. If you get a 429 out of quota error, use search_documents instead.
    ConnectorNo auth
  • AI-powered company analysis using semantic search over Nordic financial data. Orchestrates multiple searches internally and returns a synthesized narrative answer with source citations. Covers annual reports, quarterly reports, press releases and macroeconomic context for Nordic listed companies. Use this when you want a synthesized answer rather than raw search chunks. For raw data access, use search_filings or company_research instead. For a full due diligence report with AI-planned sections, use the Alfred MCP server: alfred.aidatanorge.no/mcp Args: company: Company name or ticker question: What you want to know about the company
    ConnectorNo auth
  • Create a Google Ads campaign. Defaults to channel_type=SEARCH, status=PAUSED (safe), bidding_strategy=MANUAL_CPC. Provide budget_id from google_ads_campaign_budget_create. Declares no EU political advertising by default (Reg 2024/900, required by Google). Only SEARCH, DISPLAY and SHOPPING can be created through the API: VIDEO (YouTube) is read-only on Google side, and PERFORMANCE_MAX must be created together with its asset group. Build those two in the Google Ads interface and use this MCP to report on them. Bulk support: accepts budget_ids, merchant_ids, customer_ids for batched execution.
    ConnectorNo auth
  • Run one buyer-shaped prompt through Gemini with Google Search grounding (a real, cited web search, the same mechanism behind Google AI Overviews) and report where the brand lands on the five-rung ladder: absent, cited (a page of the brand's site is a source), mentioned (named in the text), recommended (on the shortlist), or recommended against. Also returns the other names on the shortlist and the source domains the answer was built from. One prompt per call; limited to a few calls per day per user because grounded requests are billed per query.
    ConnectorNo auth