Search through the website
An assistant with browser control can use the same labelled search fields, region selector, tables and result links as you. These links fill in a search; submit it in the app when you are ready to run the research.
- Keyword research — Explore running shoes in Singapore.
- Domain research — Research example.com in Singapore.
- Page research — Research an exact page in Singapore.
Keep the selected region and exact page URL with your findings. A domain result describes the domain; it does not stand in for an individual page.
Connect an AI assistant
- Add the SearchCub MCP server URL to your assistant’s connections or MCP settings.
- Follow the browser sign-in prompt and connect your SearchCub account. Your existing browser sign-in can be reused.
- Review and approve the research access requested by your assistant. No manual token setup is required.
- MCP endpoint
https://app.searchcub.com/api/mcp- Authorization
- OAuth browser sign-in
Use an assistant that supports remote MCP with OAuth. The assistant handles connection credentials after you sign in; you do not create or paste an access token. Connection instructions are available before sign-in.
Research uses the account you connect and its existing allowance. The assistant can access only that account’s private saved results. Account and billing changes are outside these tools.
Example request: Use SearchCub to find keyword opportunities for a Singapore baking website. Start with saved or cached results and explain the sources and gaps.
Available capabilities
The API and MCP tools share this catalogue. Read-only operations return existing data or supported regions; research operations can use cached results or explicitly authorized fresh research.
- Supported research regions
list_regions - List the selectable countries for new research. Select one or several countries; Worldwide is retired. Research language is English. No new provider research.
- Keyword research
search_keywords - Find keyword suggestions, related terms, questions and autocomplete with available demand, difficulty, intent and history. Initial suggestions/all/related searches automatically research up to three selected relevant first-layer keywords per country and include up to eight new related keywords per parent. keywordBranches preserves parent reasons, child relationships and incomplete status; this second layer is included in the original keyword-search allowance and does not expand recursively. Continuation cursors cover only the original discovery streams. Optional countries selects multiple regions; additive estimates are summed and rates are volume-weighted. New research requires countries; Worldwide is retired. regionalPages preserves each country continuation. Reads current cached data by default. Fresh research needs explicit allowFresh, a fresh-enabled token and an idempotency key; normal plan search/domain allowances and provider cost limits apply. Cached-only by default; fresh research requires explicit permission.
- Domain and page keyword research
research_domain - Read rankings for a domain or exact page, or compare domain keyword gaps. Default ranking order is monthly searches descending. For one country in ranked mode, keywordView supplies keyword search, provider-supported nested filter conditions and up to two sorts before pagination. Preserve keywordView and nextOffset on later pages; restart at offset zero when criteria change. Computed grouping/history/combined-intent controls and global combined-country ordering are unavailable. Free accounts may page 100 rows at a time to a 1,000-row matching-result depth; paid accounts may request up to 1,000 rows at once. Limits apply to cached and fresh API/MCP requests. Optional countries combines selected country estimates with per-country regionalPages; these are loaded regional samples, not globally ordered complete results. Current cache only by default; fresh research needs explicit authorization and normal plan allowances. Legacy Worldwide snapshots remain readable without refresh. Cached-only by default; fresh research requires explicit permission.
- Search result snapshot
inspect_serp - Inspect a localized Google desktop organic results snapshot and available result features. A specific country is required. Current cached data by default; fresh research needs explicit authorization and the normal plan allowances. Cached-only by default; fresh research requires explicit permission.
- Domain search estimates
domain_insights - Read ranking distribution, top pages or the supplied twelve-month traffic history for a domain and country. Optional countries sums compatible selected-country metrics and retains missing-region coverage. New research requires countries; Worldwide is retired. Estimates and missing observations remain explicit. Current cached data by default; fresh research needs explicit authorization and the normal plan allowances. Cached-only by default; fresh research requires explicit permission.
- Backlinks and page research details
research_details - Read the selected backlink summary/list/history, referring domains, linked pages, page HTML checks, mobile browser performance measurements and derived page issues, page comparison or keyword visits history. Performance issues describe one lab test, not site-wide Core Web Vitals. Each operation preserves its target, coverage and activation gates. Current cached data by default; fresh research needs explicit authorization and the normal plan allowances. Cached-only by default; fresh research requires explicit permission.
- Organic competitors
organic_competitors - Read provider-known organic competitors for a domain and country. Provider index coverage is bounded. Current cached data by default; fresh research needs explicit authorization and the normal plan allowances. Cached-only by default; fresh research requires explicit permission.
- Saved research
list_research - List only the connected account's retained research with an optional offset. Never starts research, refreshes data or uses a search allowance. No new provider research.
- Saved Keywords
list_saved_keywords - List the connected account's permanently saved keywords, market, list identity and optional offset. Optional listIds (1–100 of the account's list ids; other ids match nothing) returns only those lists' keywords, and the first page reports the matching total plus, when filtered, the chosen lists that exist with their item counts. Keyword data is reused for up to 24 hours; missing or expired data stays unavailable by default. Explicit allowFresh with a fresh-enabled token and retry-stable idempotency key refreshes missing or expired keyword metrics under the keyword-search allowance, counting each refreshed keyword. Refresh failures preserve saved keywords. Never deletes keywords. Cached-only by default; fresh research requires explicit permission.
- Read or export saved research
get_research - Read one retained research run owned by the connected account as normalized JSON with source, market, dates, coverage and missing-data information. Never refreshes or uses a search allowance. No new provider research.
- Saved site audits
list_audits - List the connected account's retained site audits. Reads local evidence only and never starts or collects a provider crawl. No new provider research.
- Read a saved site audit
get_audit - Read one retained site audit owned by the connected account, preserving crawl limits, coverage, page issues and observation dates. Never starts, refreshes or collects a crawl. No new provider research.
Use the research API
Applications and agents can send JSON to POST https://app.searchcub.com/api/ai/search using the OAuth access token managed by their client. The operation selects a research capability, and query contains its inputs. The token’s resource is https://app.searchcub.com/api/mcp for both research interfaces.
curl 'https://app.searchcub.com/api/ai/search' \
-H 'Authorization: Bearer YOUR_SEARCHCUB_TOKEN' \
-H 'Content-Type: application/json' \
--data '{"operation":"search_keywords","query":{"keyword":"running shoes","country":"SG","language":"en"},"allowFresh":false}'This example allows cached research only. If the requested data is unavailable in cache, it will not start fresh research.
Read the current capability catalogue for available operations, tool descriptions and input schemas, or use the OpenAPI document when building a client. MCP tools and the API use the same research capabilities.
Read and extract results
Open Research History to read your saved research in a plain, server-rendered view or download its structured JSON. The view and export require the owning verified account. Opening saved data does not start another provider request.
Connected assistants can also retrieve saved research through the API or MCP. Responses preserve the research context and available source, freshness and coverage information, so your assistant can distinguish the result page from a complete dataset.
- Keep the keyword, domain or exact page, selected market and language with each result.
- Treat search volume, difficulty, traffic and other provider estimates as estimates.
- Keep missing values and unavailable source states distinct from genuine zero values.
- Follow returned pagination and coverage information instead of assuming the loaded rows are exhaustive.
- Use the saved result link and its dates when referring to a research snapshot.
Fresh research and access
Fresh research requires the research permission you approve during connection, an explicit allowFresh: true request and an idempotencyKey. Use a new key for each intended fresh action and reuse it when retrying that action. Normal account permissions, credits and usage limits still apply.
Saved reads and cached-only requests do not grant permission to run a new search. Expired or unavailable data remains explicit, and private account data is not published for search engines or other users.
Capabilities, their schemas and this guidance are maintained alongside SearchCub’s product changes. Public guides explain how to interpret the research, and pricing describes the current plans.