Visibility Genies
Server Details
Your AI assistant reads and runs SEO and AI visibility work in your Visibility Genies account.
- Status
- Healthy
- Uptime
- 100.0% over the last 24 hours
- OAuth
- Works in Glama
- Last Tested
- Transport
- Streamable HTTP · MCP 2025-11-25
- URL
- ChatGPT
- –
- Claude
- –
TDQS
Scored across 158 tools
The descriptions are unusually detailed and cross-reference each other, but with 158 tools there are many near-neighbors an agent can easily confuse: live purchases versus get_saved_* readers, get_backlinks_overview/profile/report, get_google_analytics_traffic_breakdown/acquisition/audience_breakdown, and lookup_brand versus get_brand_mentions/get_brand_performance. The saved-versus-live split is logical but creates dozens of paired tools whose boundaries require careful reading.
Most tools follow a clear snake_case verb_noun pattern (get_, list_, create_, save_, find_, run_, update_), and the naming is largely predictable. Minor deviations exist: whoami, lookup_brand uses a different verb style than get_brand, and singular/plural pairs like save_competitor/save_competitors can cause small confusion.
158 tools is an extreme mismatch for a single MCP server and far above the 3-15 range that keeps a surface navigable. While the platform genuinely spans SEO, analytics, ads, local, reviews, content and AI prompts, many tools could be consolidated (especially saved-versus-live pairs and per-platform ad readers).
Coverage is very broad: projects and clients, audits, rank tracking, backlinks, keyword research, SERPs, AI prompt tracking, brand/AI mention analysis, local and review data, ad platforms, Google Analytics/Search Console, content templates, drafts, knowledge base, actions and spend. Gaps are minor but real: keyword strategy, content briefs and prompt tracking setup are page-only, projects support archive/restore but not permanent delete, and disavow management is read-only.
Available Tools
158 toolsadd_action_noteAdd a note to an actionAInspect
Add a note or comment to a saved action, shown with the action as added by your agent. Notes are kept; they cannot be edited or removed here. Uses no credits or provider calls.
Inputs: id (required), text (required), projectId (required). Returns: meta, note
add_rank_tracking_keywordsAdd rank tracking keywordsAInspect
Add keywords to an existing rank tracker. This buys nothing and starts no check. On a scheduled tracker first show the user the price from estimate_rank_tracker_cost and pass that figure as maxEstimatedScheduledCheckCredits. A scheduled tracker buys its checks only under the user's approval of its settings, given on the tracker's page: without one its schedule is paused, buying nothing, until they approve there (scheduleAwaitsApproval). Keywords already tracked and repeats are skipped; added is the number really added.
Inputs: keywords (required), projectId (required), trackerId (required), matchCase, +1 more. Returns: meta, added, addedIds, requested, +2 more
archive_projectArchive a projectADestructiveInspect
Archives an active project, keeping its saved data so restore_project can bring it back. Free: buys no data. Use a projectId from list_projects. The workspace's only active project cannot be archived. This does not permanently delete anything.
Inputs: projectId (required). Returns: success, projectId
ask_ai_modelsAsk AI modelsAInspect
Asks one prompt of the ChatGPT, Claude, Gemini and Perplexity APIs (and Grok, DeepSeek and Muse Spark, only when named; the last two answer from what the model learned, without searching the web) and the ChatGPT and Gemini apps, as a Run on Prompt Explorer does: each answer and the sources it cites, the searches the ChatGPT app ran, and whether each names a brand when one is given. Estimated price: $0.01 to $1.06 an answer from an API model, by model and version, and $0.01 an answer from an app. An answer saved for the same prompt, model, version, web search and country, here or on the page, is read free and not bought again; only answers with nothing saved are bought. Only an app's answer is ever bought again, once out of date or empty, and only with refreshApps; for fresh answers from the API models over time, track the prompt on the Prompt Tracking page, which buys new ones on a schedule. Answers are kept in Prompt Explorer's Saved answers and History; get_saved_prompt_answers reads them without buying. Several answers can take a few minutes; if your client gives up first, the answers that finished are still saved, so read them with get_saved_prompt_answers before calling again and buying them twice. Each call may spend up to the workspace's agent spend limit ($2.00 unless changed in Settings) without asking: it holds at most its estimate plus a margin while it runs and is charged what it bought, never more. A call whose maximum is above the limit is refused before anything is bought, with its estimate and maximum: ask the user, then call again with approvedMaxCostUsd set to the amount they approved.
Inputs: prompt (required), projectId (required), apps, brand, +6 more. Returns: cost, meta, bought, prompt, +3 more
build_content_templateBuild a content templateAInspect
Builds a keyword's content template, from its top 10 Google results: the typical length of the pages (the middle half of their word counts, a fact and not a target: Google sets no word count), the median number of subheadings and the median readability (0 hard to 100 easy), which help readers and are not Google ranking factors, the terms most of the pages use, each in all its forms, with how many use each, and each page with its headings, or why it could not be read. Length and terms are compared only when at least 5 pages could be read. Estimated price per keyword: up to $0.02, for the SERP analysis, when none of the last 12 hours is saved, then each page with no saved copy. A template whose results are still current is answered from what is saved at no charge, and every build opens free on the Content Template page. Pages that could not be bought are listed, and the next call buys only them. When Google hid the address of every top result on desktop, the mobile results are bought once more, charged on top of that estimate but never above the call's maximum, and the template says it came from mobile. When Google hid the address of some results, the template is built from the others (at least 3 must be readable) and names the ones it could not read; with fewer, no template is built and the answer says what the search results were charged. Each call may spend up to the workspace's agent spend limit ($2.00 unless changed in Settings) without asking: it holds at most its estimate plus a margin while it runs and is charged what it bought, never more. A call whose maximum is above the limit is refused before anything is bought, with its estimate and maximum: ask the user, then call again with approvedMaxCostUsd set to the amount they approved.
Inputs: keyword (required), projectId (required), mobile, languageCode, +2 more. Returns: cost, meta, bought, savedAt, +3 more
cancel_competitor_watchStop a competitor watchAInspect
Cancel future monthly competitor purchases for the saved watch. No provider calls. Already completed purchases remain saved.
Inputs: id (required), projectId (required). Returns: meta, result
check_page_seoCheck a pageAInspect
Runs an On Page SEO check: ideas for one page of the project and the keyword it should rank for, from comparing it with the top 10 Google results: content (title, main heading, meta description, indexing), terms to cover (every term most top pages use that the page misses, in one row), internal links (the links to it from its own site, from Site Audit), backlinks (referring domains beside the top pages') and Core Web Vitals, each with the numbers behind it, ranked in that order of what matters for search, a missing meta description last since it does not affect ranking; then, for readers and not numbered with the ideas, since they are not Google ranking factors, its length, headings and readability beside the top pages', as facts; and the page's match with the top pages out of 100, the share of the keyword's content template terms it covers. Estimated price per On Page SEO check: up to $0.08, for the keyword's content template when no current one is saved, the page's text when none is saved and Visibility Genies' crawler, which reads it free first, cannot read it, and the referring domains of the page and the top pages with none saved in the last 24 hours. Title and indexing come from what Site Audit saved, and Core Web Vitals from real Chrome users (the free Chrome UX Report) when the workspace has a key and the site enough visits, else Site Audit's lab Lighthouse test; none is bought. An On Page SEO check whose data is all current is answered from what is saved at no charge, and every one opens free on the On Page SEO Checker page. With readPageAgain, it reads only the page's text again, free by Visibility Genies' crawler first and bought only when it cannot, for up to $0.01, and saves a new check from it. Each call may spend up to the workspace's agent spend limit ($2.00 unless changed in Settings) without asking: it holds at most its estimate plus a margin while it runs and is charged what it bought, never more. A call whose maximum is above the limit is refused before anything is bought, with its estimate and maximum: ask the user, then call again with approvedMaxCostUsd set to the amount they approved.
Inputs: keyword (required), pageUrl (required), projectId (required), languageCode, +3 more. Returns: cost, meta, check, bought, +3 more
compare_brandsCompare brandsAInspect
Brand Lookup's Competitor Research: a brand and 1 to 5 competitors side by side in AI answers (United States, English; ChatGPT and Google AI Overview): each one's mentions and AI search volume on each platform, and its Share of Voice. Estimated price: up to $0.44 for 5 competitors, up to $0.44 for 1: one AI mentions request per platform for every brand. A comparison saved within 24 hours, here or on the Brand Lookup page, is read free. It opens free on Brand Lookup; get_saved_brand_comparison reads it without buying. Each call may spend up to the workspace's agent spend limit ($2.00 unless changed in Settings) without asking: it holds at most its estimate plus a margin while it runs and is charged what it bought, never more. A call whose maximum is above the limit is refused before anything is bought, with its estimate and maximum: ask the user, then call again with approvedMaxCostUsd set to the amount they approved.
Inputs: query (required), projectId (required), competitors (required), names, +2 more. Returns: cost, meta, query, saved, +3 more
compare_domainsCompare domainsAInspect
Puts 2 to 5 domains side by side in one market: organic traffic and keywords (each domain's Domain Overview totals), paid search traffic and keywords, and each domain's ranking changes (keywords new, up, down and lost in the last check). Estimated price: up to $0.15 for 5 domains, up to $0.08 for 2: one request for the traffic of all of them and one for each domain's totals. Totals saved within 12 hours, here, on Domain Overview or by get_domain_overview, are read free, and so is a comparison saved within 12 hours. It opens free on the Compare Domains page; get_saved_domain_comparison reads it without buying. Each call may spend up to the workspace's agent spend limit ($2.00 unless changed in Settings) without asking: it holds at most its estimate plus a margin while it runs and is charged what it bought, never more. A call whose maximum is above the limit is refused before anything is bought, with its estimate and maximum: ask the user, then call again with approvedMaxCostUsd set to the amount they approved.
Inputs: domains (required), projectId (required), languageCode, locationCode, +1 more. Returns: cost, meta, rows, saved, +2 more
confirm_competitor_refreshApprove a competitor quoteAInspect
Only call after the user approves this exact quote, payer and price. Submit its unmodified id and confirmationHash. Refresh confirmation records approval; execute_competitor_refresh starts one component. A monthly watch cannot be confirmed here: it buys every 30 days only under the Owner's approval, which they give on the Competitors page, so show the user the watch quote and send them there. Changed or expired quotes require a fresh quote and approval.
Inputs: id (required), projectId (required), confirmationHash (required), kind. Returns: meta, result
create_actionCreate an actionAInspect
Adds one action to the project's Actions page, in Suggested, labelled as added by your agent: a title, why it matters, the steps and the pages it affects. Ask the user before adding actions. Each call adds one action; at most 20 a project each day. Refreshes never change or reopen it, and the user moves it to To do, done or dismissed. Free: buys nothing. Returns the action's id for update_action.
Inputs: why (required), steps (required), title (required), projectId (required), +1 more. Returns: meta, action
create_projectCreate a projectAInspect
Create a new project in the user's organization. Free: buys no data. Provide a name, and optionally a domain, a default market (locationCode/languageCode; a languageCode requires a locationCode) and the client to group it under (clientId from list_clients, or newClientName). Returns the created {id, name, domain, locationCode, languageCode, url} and its client; pass the returned id as projectId to other Visibility Genies tools. Call list_projects first to avoid creating a duplicate.
Inputs: name (required), domain, clientId, languageCode, +3 more. Returns: meta, project
create_rank_trackerCreate a rank trackerAInspect
Create a rank tracking configuration for a project. Creating an empty tracker buys nothing and starts no check. A daily, weekly or monthly tracker buys its checks once it has keywords, and only after the user approves its price on the tracker's page. The domain defaults to the project's domain; market defaults to the project's market; devices default to mobile, search depth to 40, AI Overviews to checked, and schedule to manual; re-adding an archived tracker keeps its own devices, depth and AI Overviews unless they are given, and brings back its keywords: the reply prices them, and a scheduled one stays paused, buying nothing, until the user approves that price on the tracker's page. Use estimate_rank_tracker_cost before adding keywords to a scheduled tracker or starting a live run. Call get_rank_tracker first to avoid duplicates.
Inputs: projectId (required), domain, devices, serpDepth, +5 more. Returns: meta, config, trackerId
dismiss_actionDismiss an actionADestructiveInspect
Move a saved action to Dismissed. Nothing is deleted: restore_action puts it back. A dismissed audit finding returns only if more pages are affected or it gets worse; other actions stay dismissed until restored. Uses no credits or provider calls.
Inputs: id (required), projectId (required). Returns: meta, action
edit_actionEdit an actionAInspect
Change a saved action's title, why it matters, steps, pages or due day. Only the fields you give change; the others stay as saved. Ask the user before changing their actions. The change is logged on the action as made by your agent, and refreshes never overwrite it. Uses no credits or provider calls.
Inputs: id (required), projectId (required), why, dueOn, +3 more. Returns: meta, action
estimate_rank_tracker_costEstimate rank tracking costARead-onlyInspect
Estimate what a tracker costs without buying anything or starting a check. The live figure covers one explicit run_rank_tracker check. For a scheduled tracker the reply also gives the most each queued check, and a month of them, can cost. Pass additionalKeywordCount before adding keywords to see the cost after they are added. A scheduled check never costs more than that figure; a live check only replaces a queued one that was never posted, within what that post would have cost.
Inputs: projectId (required), trackerId (required), additionalKeywordCount. Returns: meta, method, costUsd, trackerId, +8 more
execute_competitor_refreshRun a competitor refresh stepAInspect
May spend money. Execute the next component of a previously confirmed quote. Never call without the user's exact approval. Read returned progress before continuing. Do not retry uncertain or unreadable work; use recover_competitor_refresh to reconcile without buying again. Server enforces pinned payer, payload and price. Each call may spend up to the workspace's agent spend limit ($2.00 unless changed in Settings) without asking: it holds at most its estimate plus a margin while it runs and is charged what it bought, never more. A call whose maximum is above the limit is refused before anything is bought, with its estimate and maximum: ask the user, then call again with approvedMaxCostUsd set to the amount they approved.
Inputs: id (required), projectId (required), approvedMaxCostUsd. Returns: cost, meta, result, savedResultId
find_ad_advertisersFind advertisersAInspect
Finds the advertisers Google's Ads Transparency Center lists under a name or domain, in the project's country, with each one's advertiserId and approximate ad count. Pass an advertiserId (or a domain) to get_competitor_ads. Queued at normal priority: a 'processing' answer means call again with the same arguments in 2-3 minutes, at no extra cost. Estimated price: $0.01 a search. Each call may spend up to the workspace's agent spend limit ($2.00 unless changed in Settings) without asking: it holds at most its estimate plus a margin while it runs and is charged what it bought, never more. A call whose maximum is above the limit is refused before anything is bought, with its estimate and maximum: ask the user, then call again with approvedMaxCostUsd set to the amount they approved.
Inputs: query (required), projectId (required), approvedMaxCostUsd. Returns: cost, meta, status, advertisers, +1 more
find_ai_cited_sitesFind sites AI answers citeAInspect
Finds the 50 sites ChatGPT's answers about a topic cite as sources most (United States, English), as link prospects: keep one with save_link_prospect. Estimated price: up to $0.30, billed by the rows returned. An answer saved for the topic within a day is read free. Each call may spend up to the workspace's agent spend limit ($2.00 unless changed in Settings) without asking: it holds at most its estimate plus a margin while it runs and is charged what it bought, never more. A call whose maximum is above the limit is refused before anything is bought, with its estimate and maximum: ask the user, then call again with approvedMaxCostUsd set to the amount they approved.
Inputs: projectId (required), topic, approvedMaxCostUsd. Returns: cost, meta, rows, bought, +1 more
find_link_competitorsFind link competitorsAInspect
Finds the 20 sites that share the most linking sites with a domain: rivals for the same links, to track as competitors (save_competitor) or compare in get_backlink_gap. Estimated price: up to $0.05, one backlinks request billed by the rows returned. An answer saved for the domain within 7 days is read free. Each call may spend up to the workspace's agent spend limit ($2.00 unless changed in Settings) without asking: it holds at most its estimate plus a margin while it runs and is charged what it bought, never more. A call whose maximum is above the limit is refused before anything is bought, with its estimate and maximum: ask the user, then call again with approvedMaxCostUsd set to the amount they approved.
Inputs: projectId (required), domain, approvedMaxCostUsd. Returns: cost, meta, rows, bought, +1 more
find_locationFind a location codeARead-onlyInspect
Looks up the numeric location code a place has, instead of guessing one. Countries are the markets keyword, domain and backlink tools take as locationCode; with countryCode, cities and regions of that country are looked up too, for search results and rank trackers. A misspelled name returns the nearest names. Free: it reads the saved location lists and never fetches anything; a country whose city list is not saved yet says so.
Inputs: query (required), countryCode. Returns: exact, places, countries, placesSaved
find_questionsFind questions people askAInspect
Buys Questions: up to 200 questions people search about a topic (found by question words in the market's language: English, German, French, Spanish, Italian, Dutch and Portuguese have their own, and other languages use the English ones), most searched first, each with Google search volume, difficulty, AI search volume, and the People Also Ask question on the topic's saved SERP analysis it rewords (about 70% of its words shared), with each People Also Ask question listed. Estimated price up to $0.14 per topic: the questions, then AI search volume only for questions with none saved. Questions bought in the last 24 hours, and AI search volume already bought for a keyword in this project and market, are read at no charge, and every purchase opens free on the Questions page. The SERP is only read, never bought. Each call may spend up to the workspace's agent spend limit ($2.00 unless changed in Settings) without asking: it holds at most its estimate plus a margin while it runs and is charged what it bought, never more. A call whose maximum is above the limit is refused before anything is bought, with its estimate and maximum: ask the user, then call again with approvedMaxCostUsd set to the amount they approved.
Inputs: topic (required), projectId (required), languageCode, locationCode, +1 more. Returns: cost, meta, bought, savedAt, +3 more
find_saved_competitor_suggestionsFind competitor suggestionsBInspect
Find suggestions from saved evidence only. Saves suggestions without accepting them. Follow returned coverage and cursor until complete; ignored rivals stay ignored. No provider calls or credits.
Inputs: projectId (required), cursor, market, status, +1 more. Returns: meta, result
find_serp_competitorsFind search competitorsAInspect
Compares the domains that compete in Google results for a set of keywords, in a country-level market (location and language codes). It is not radius based local SEO. Returns each domain's keyword count, average and median position, visibility and estimated traffic. Estimated price: ~$0.04 for 50 domains, billed by the domains returned; the same keywords, market and options asked again within 12 hours are answered from the saved copy at no charge. Each call may spend up to the workspace's agent spend limit ($2.00 unless changed in Settings) without asking: it holds at most its estimate plus a margin while it runs and is charged what it bought, never more. A call whose maximum is above the limit is refused before anything is bought, with its estimate and maximum: ask the user, then call again with approvedMaxCostUsd set to the amount they approved.
Inputs: keywords (required), projectId (required), limit, market, +8 more. Returns: cost, meta, competitors, savedResultId
find_sites_using_technologyFind sites using a technologyAInspect
A reverse tech stack lookup for prospect lists: the strongest sites in one country that run a technology (a CMS, shop platform, analytics or other tool), each with its title, domain rank (a 0 to 1,000 link-graph score), language and when the crawler last visited it. Estimated price: up to $0.27, one request for up to 100 sites, billed by the sites returned. The same technology and country saved within 30 days is read free, and it opens free on Domain Overview's Tech stack tab. Each call may spend up to the workspace's agent spend limit ($2.00 unless changed in Settings) without asking: it holds at most its estimate plus a margin while it runs and is charged what it bought, never more. A call whose maximum is above the limit is refused before anything is bought, with its estimate and maximum: ask the user, then call again with approvedMaxCostUsd set to the amount they approved.
Inputs: projectId (required), technology (required), countryCode (required), approvedMaxCostUsd. Returns: cost, meta, bought, domains, +1 more
find_top_searchesFind top searchesAInspect
The most searched keywords in a market, highest monthly volume first, narrowed by text they contain, a least volume and a highest keyword difficulty, each with its volume, CPC, difficulty and search intent. Estimated price: up to $0.05, one market data request for up to 100 keywords, billed by the rows returned. The same filters and market saved within a day are read free, and they open free on the Top searches page. Each call may spend up to the workspace's agent spend limit ($2.00 unless changed in Settings) without asking: it holds at most its estimate plus a margin while it runs and is charged what it bought, never more. A call whose maximum is above the limit is refused before anything is bought, with its estimate and maximum: ask the user, then call again with approvedMaxCostUsd set to the amount they approved.
Inputs: projectId (required), contains, minVolume, languageCode, +3 more. Returns: cost, meta, bought, keywords, +1 more
get_actionRead an actionARead-onlyInspect
Read one saved action with its steps, pages, due day, notes and change history, each change saying whether it was made on the Actions page, by your agent or via Genie. Uses no credits or provider calls.
Inputs: id (required), projectId (required). Returns: meta, notes, action, changes
get_ai_crawler_visitsRead AI crawler visitsARead-onlyInspect
Reads the AI crawler visits saved for a project: hits per day per crawler (ChatGPT, Claude, Perplexity, Common Crawl and others), split into verified (the address is in the list the crawler's operator publishes) and name only, the requests the site refused (401, 402, 403 or 429), each crawler's last visit, and its most fetched pages. Visits come from the project's connected Cloudflare zone, read every day, or from server logs uploaded on the AI crawler visits page; source picks one, and the default is Cloudflare when it holds visits. Free: reads saved data. Connecting and uploading happen on the page only.
Inputs: projectId (required), days, source. Returns: meta, counts, source, crawlers, +1 more
get_ai_mention_changesGet AI mentions gained and lostAInspect
Buys the AI mentions a brand or domain gained and lost for this project, as Brand Lookup does: month by month over the last year, how many ChatGPT and Google AI Overview answers started mentioning it, how many stopped, and the net change, in mentions and AI search volume, in the United States. Estimated price: ~$0.92 for both platforms. What is saved and current (for 24 hours), here or on the page, is read free, and only a platform missing or out of date is bought. It opens free on Brand Lookup. Each call may spend up to the workspace's agent spend limit ($2.00 unless changed in Settings) without asking: it holds at most its estimate plus a margin while it runs and is charged what it bought, never more. A call whose maximum is above the limit is refused before anything is bought, with its estimate and maximum: ask the user, then call again with approvedMaxCostUsd set to the amount they approved.
Inputs: query (required), projectId (required), scope, approvedMaxCostUsd. Returns: cost, meta, bought, platforms, +2 more
get_ai_mentions_over_timeGet AI mentions over timeAInspect
Buys the AI mentions over time of a brand or domain for this project, as Brand Lookup does: its mentions and AI search volume in ChatGPT and Google's AI Overviews, month by month, in the United States. Estimated price: ~$0.44 for both platforms. What is saved and current (for 24 hours), here or on the page, is read free, and only a platform missing or out of date is bought. It opens free on Brand Lookup; get_saved_ai_mentions_over_time reads it without buying. Each call may spend up to the workspace's agent spend limit ($2.00 unless changed in Settings) without asking: it holds at most its estimate plus a margin while it runs and is charged what it bought, never more. A call whose maximum is above the limit is refused before anything is bought, with its estimate and maximum: ask the user, then call again with approvedMaxCostUsd set to the amount they approved.
Inputs: query (required), projectId (required), scope, approvedMaxCostUsd. Returns: cost, meta, bought, platforms, +2 more
get_audit_issuesRead audit issuesARead-onlyInspect
Reads the findings of a site audit, most serious first, with a count per check, what each check means and how to fix it. Free: reads saved audit data. Findings and checks the project hid in Site Audit are left out, and the answer says how many. Omit auditId for the audit Site Audit shows.
Inputs: projectId (required), limit, auditId, severity, +1 more. Returns: meta, issues, summary
get_audit_pagesRead audit pagesARead-onlyInspect
Lists the pages a site audit crawled with status, fetch result (with the reason in failText when a page could not be loaded), title, size and link counts, plus Lighthouse scores for the pages it tested. Free: reads saved audit data. Lighthouse scores come from one lab test per device, not from real visitors, and INP is not measured, so Total Blocking Time stands in for it. Omit auditId for the audit Site Audit shows.
Inputs: projectId (required), limit, auditId, fetchClass, +3 more. Returns: meta, pages, total
get_audit_statusRead audit statusAInspect
Shows how a site audit is going: its phase, pages crawled and Lighthouse progress, which crawler identity it used (crawlerIdentity), its pace (fasterCrawl is true only for the Owner's faster crawl of their own server; otherwise about a page a second), the addresses the crawler sends from (crawlerAddresses), whether it rendered pages with JavaScript (renderJavaScript; a page that could not be rendered was read as plain HTML) and whether the Owner's setting let it ignore robots.txt (robotsTxtIgnored). A finished audit also reports its Site health and AI readiness scores out of 100, matching Site Audit. Free: it reads Visibility Genies data, and it may mark an audit failed when its workflow has died. Leave out auditId for the latest audit.
Inputs: projectId (required), auditId. Returns: meta, status
get_backlink_gapGet the backlink gapAInspect
Finds link building prospects: the sites that link to up to 4 competitors and not to your domain, each with the competitors it links to (at least, when a competitor has more referring domains than its top rows saved), whether it looks sitewide (more than 50 backlinks to one competitor), the highest rank its links pass (rank passed, a 0 to 100 link-graph score, not a Google metric, which sitewide links inflate), its average link spam and its backlinks to them. Estimated price: up to $0.12 for each competitor, one backlinks request for its top 1,000 referring domains by rank, billed by the rows returned. A competitor saved for this domain within 6 hours is read free, and the gap opens free on the Backlink Gap page; get_saved_backlink_gap reads it without buying. Each call may spend up to the workspace's agent spend limit ($2.00 unless changed in Settings) without asking: it holds at most its estimate plus a margin while it runs and is charged what it bought, never more. A call whose maximum is above the limit is refused before anything is bought, with its estimate and maximum: ask the user, then call again with approvedMaxCostUsd set to the amount they approved.
Inputs: projectId (required), competitors (required), limit, domain, +2 more. Returns: cost, meta, rows, saved, +3 more
get_backlink_gap_pagesGet the backlink gap by pageAInspect
Finds link building prospects page by page: up to 500 referring pages that link to every competitor named and not to your domain, strongest link first, with each page's title, domain rank, link spam and its anchor to each competitor. Use get_backlink_gap for the sites instead. Estimated price: up to $0.09 for one competitor, up to ~$0.20 for 4, one backlinks request billed by the rows returned. The same set saved within 6 hours is read free, and it opens free on the Backlink Gap page, By page. Each call may spend up to the workspace's agent spend limit ($2.00 unless changed in Settings) without asking: it holds at most its estimate plus a margin while it runs and is charged what it bought, never more. A call whose maximum is above the limit is refused before anything is bought, with its estimate and maximum: ask the user, then call again with approvedMaxCostUsd set to the amount they approved.
Inputs: projectId (required), competitors (required), domain, approvedMaxCostUsd. Returns: cost, meta, rows, domain, +2 more
get_backlinks_overviewGet a backlinks overviewAInspect
Returns a backlinks profile summary (total backlinks, referring domains, domain rank, spam scores, 12 months of trends) and the top referring domains, each with the rank its links pass to the target (sitewide links inflate it). Ranks and spam scores are 0 to 100 scores, not Google metrics. Estimated price: ~$0.16 for a domain (three Backlinks API calls: totals, 12 months of trends, 100 referring domains), ~$0.10 without the referring domains, ~$0.11 for a single page. These are the Backlinks page's own purchases: what is bought here opens free there, and what it holds is answered here at no charge while it is current. Bare domains default to scope 'subdomains'; pass scope 'domain' to exclude subdomains from the totals. Targets with a path default to 'subfolder', whose counts come from filtered backlink totals (no rank, trends or referring-domain breakdown). Trend data always includes subdomains (provider limitation). Each call may spend up to the workspace's agent spend limit ($2.00 unless changed in Settings) without asking: it holds at most its estimate plus a margin while it runs and is charged what it bought, never more. A call whose maximum is above the limit is refused before anything is bought, with its estimate and maximum: ask the user, then call again with approvedMaxCostUsd set to the amount they approved.
Inputs: target (required), projectId (required), scope, hideSpam, +2 more. Returns: cost, meta, scope, target, +4 more
get_backlinks_profileGet a backlinks profileAInspect
Returns one page of backlink rows for a domain or URL, with filters, sorting, a mode and paging. Every link is live; lost links are not bought. Estimated price: ~$0.06 for 50 rows, ~$0.07 for 200, billed by the rows returned. It is the Backlinks page's own table: a page bought here opens free there and the other way round, while its saved copy is current. Each call may spend up to the workspace's agent spend limit ($2.00 unless changed in Settings) without asking: it holds at most its estimate plus a margin while it runs and is charged what it bought, never more. A call whose maximum is above the limit is refused before anything is bought, with its estimate and maximum: ask the user, then call again with approvedMaxCostUsd set to the amount they approved.
Inputs: target (required), projectId (required), mode, page, +7 more. Returns: cost, meta, scope, target, +2 more
get_backlinks_reportGet a backlinks reportAInspect
Buys one Backlinks report for a domain or page: best_pages (up to 100 of a domain's pages with the most linking sites, with titles) or referring_networks (up to 100 IP addresses or subnets its links come from). Estimated price: up to ~$0.06 a report, one backlinks request billed by the rows returned. A report saved within 6 hours is read free, and it opens free on the Backlinks page's tab of the same name. For the sites whose links overlap a domain's, use find_link_competitors. Each call may spend up to the workspace's agent spend limit ($2.00 unless changed in Settings) without asking: it holds at most its estimate plus a margin while it runs and is charged what it bought, never more. A call whose maximum is above the limit is refused before anything is bought, with its estimate and maximum: ask the user, then call again with approvedMaxCostUsd set to the amount they approved.
Inputs: report (required), target (required), projectId (required), scope, +2 more. Returns: cost, meta, rows, report, +2 more
get_brand_mentionsGet brand mentionsAInspect
Brand Mentions' tone and trend for a brand on the web (the data service's content analysis), the project's own site left out: the sites that write about it most positively, negatively and neutrally, how often it is mentioned month by month for 12 months, and the spread of ratings on pages that rate it. Estimated price up to $0.18, billed by the rows returned. The web pages that name a brand, and their totals, are get_web_mentions'. Mentions the Brand Mentions page bought are listed too, free. A part saved within 24 hours, here or on the page, is read free; get_saved_brand_mentions reads it without buying. Each call may spend up to the workspace's agent spend limit ($2.00 unless changed in Settings) without asking: it holds at most its estimate plus a margin while it runs and is charged what it bought, never more. A call whose maximum is above the limit is refused before anything is bought, with its estimate and maximum: ask the user, then call again with approvedMaxCostUsd set to the amount they approved.
Inputs: projectId (required), brand, excludeDomain, approvedMaxCostUsd. Returns: cost, meta, saved, keyword, +4 more
get_brand_performanceGet brand performanceAInspect
Brand Performance: for a topic or keyword, the up to 100 brands ChatGPT's answers mention most, with their mentions, AI search volume and Share of Voice, the brand categories they mention most, and where a brand stands among them (United States, English; ChatGPT: the only market and platform with brand data). Estimated price: up to $0.64, one AI mentions request for the brands and one for the categories, billed by the rows they return. A topic saved within 24 hours, here or on the Brand Performance page, is read free. It opens free on that page; get_saved_brand_performance reads it without buying. Each call may spend up to the workspace's agent spend limit ($2.00 unless changed in Settings) without asking: it holds at most its estimate plus a margin while it runs and is charged what it bought, never more. A call whose maximum is above the limit is refused before anything is bought, with its estimate and maximum: ask the user, then call again with approvedMaxCostUsd set to the amount they approved.
Inputs: topic (required), projectId (required), brand, approvedMaxCostUsd. Returns: cost, meta, saved, topic, +3 more
get_business_profileGet a business profileAInspect
Reads one business's Google Business Profile: category, rating and its breakdown, address, phone, website, whether it is claimed, opening hours, photo count and Google's cid and place_id. Name the business with exactly one of businessName, cid or placeId; use near when a name is ambiguous. Estimated price: $0.02 per profile; a profile bought within a day, here or on the Local page, is answered at no charge. Each call may spend up to the workspace's agent spend limit ($2.00 unless changed in Settings) without asking: it holds at most its estimate plus a margin while it runs and is charged what it bought, never more. A call whose maximum is above the limit is refused before anything is bought, with its estimate and maximum: ask the user, then call again with approvedMaxCostUsd set to the amount they approved.
Inputs: projectId (required), cid, near, placeId, +4 more. Returns: cost, meta, profile, savedResultId
get_business_reviewsGet business reviewsAInspect
Collects the Google reviews of one business, with rating, author, date, text and the owner's reply, for review-gap analysis. Name the business with exactly one of businessName, cid or placeId (cid is the most precise). Usually completes within this call; if the queued job is still running you get status 'processing' plus a taskId, so call again with that taskId in 30-60 seconds to collect it at no extra cost. Estimated price: $0.01 for 20 reviews, billed per 10 (per 20 with includeOtherSources) at twice the Local page's normal-priority rate, so the call usually returns them; reviews bought within a day, here or on the Local page, are answered at no charge. Each call may spend up to the workspace's agent spend limit ($2.00 unless changed in Settings) without asking: it holds at most its estimate plus a margin while it runs and is charged what it bought, never more. A call whose maximum is above the limit is refused before anything is bought, with its estimate and maximum: ask the user, then call again with approvedMaxCostUsd set to the amount they approved.
Inputs: projectId (required), cid, near, depth, +8 more. Returns: cost, meta, status, taskId, +3 more
get_business_updatesGet business postsAInspect
Collects the posts, offers and events a business published on its Google Business Profile, newest first. Name the business with exactly one of businessName, cid or placeId. Usually completes within this call; if the queued job is still running you get status 'processing' plus a taskId, so call again with that taskId in 30-60 seconds to collect it at no extra cost. Estimated price: $0.01 for 10 posts, at twice the Local page's normal-priority rate; posts bought within a day, here or on the Local page, are answered at no charge. Each call may spend up to the workspace's agent spend limit ($2.00 unless changed in Settings) without asking: it holds at most its estimate plus a margin while it runs and is charged what it bought, never more. A call whose maximum is above the limit is refused before anything is bought, with its estimate and maximum: ask the user, then call again with approvedMaxCostUsd set to the amount they approved.
Inputs: projectId (required), cid, near, depth, +6 more. Returns: cost, meta, status, taskId, +2 more
get_chatgpt_ads_performanceRead ChatGPT Ads performanceARead-onlyInspect
The connected ChatGPT Ads account's spend, impressions, clicks, CTR, CPC, conversions, cost per conversion, attributed sales and ROAS for a date range, in total and per campaign, most spend first, from what the daily pull stored. Conversions and sales counted within 30 days of a click or 1 day of a view, as ChatGPT Ads Manager counts them by default. Money is in the account's currency (currencyCode). Days are in the account's time zone; at most 90 days. ChatGPT Ads shares no prompts or topics. Read-only; uses no credits.
Inputs: projectId (required), endDate, startDate. Returns: ok, meta, reason, totals, +10 more
get_client_reportRead the client reportARead-onlyInspect
The project's client report as its client would read it, the same one Project settings previews: the Summary's five answers, the scores and what changed, the actions done so far this calendar month (period, in UTC) and the next ones, and the ad results where an ad account is connected, as of today. Read it to write a monthly report; send nothing to a client without the user's review. Creates no link: create_report_share does that, only when the user asks. Free: it reads saved data and buys nothing.
Inputs: projectId (required). Returns: asOf, meta, period, report
get_competitor_adsGet a competitor's adsAInspect
Collects the ads one advertiser (or a domain) ran on Google Search, YouTube, Shopping, Maps and Play, in the project's country, from Google's Ads Transparency Center: each ad's format, when it was first and last shown, and a preview URL. Queued at normal priority: a 'processing' answer means call again with the same arguments in 2-3 minutes, at no extra cost. Estimated price: $0.01 for 40 ads, billed per 40. Each call may spend up to the workspace's agent spend limit ($2.00 unless changed in Settings) without asking: it holds at most its estimate plus a margin while it runs and is charged what it bought, never more. A call whose maximum is above the limit is refused before anything is bought, with its estimate and maximum: ask the user, then call again with approvedMaxCostUsd set to the amount they approved.
Inputs: projectId (required), depth, domain, platform, +2 more. Returns: ads, cost, meta, status, +1 more
get_competitor_link_changesGet competitor link changesAInspect
Counts the linking sites the project and each tracked competitor gained and lost in the last 30 days, all in one request. Estimated price: up to up to $0.06, one backlinks request plus a row per site. A count saved within a day that holds every site asked is read free. Each call may spend up to the workspace's agent spend limit ($2.00 unless changed in Settings) without asking: it holds at most its estimate plus a margin while it runs and is charged what it bought, never more. A call whose maximum is above the limit is refused before anything is bought, with its estimate and maximum: ask the user, then call again with approvedMaxCostUsd set to the amount they approved.
Inputs: projectId (required), sites, approvedMaxCostUsd. Returns: cost, meta, rows, bought, +1 more
get_disavow_listRead the disavow listARead-onlyInspect
Reads the sites and links the project marked to disavow on the Backlinks page, each with when it was marked, and the disavow file they make for Google Search Console. Read only: no agent tool marks or unmarks one. Uploading the file replaces the list Search Console holds. Uses no credits and calls no provider.
Inputs: projectId (required). Returns: meta, links, sites, fileText
get_domain_overviewGet a domain overviewAInspect
Returns a domain's organic footprint in one market: estimated organic traffic and how many keywords it ranks for. Use this first for domain research, then get_ranked_keywords for the keywords themselves, and get_backlinks_overview for links. Estimated price: ~$0.03, one data request. It is Domain Overview's totals: a domain bought here opens free there and the other way round, and asking again within 12 hours is answered from the saved copy at no charge, an empty answer included. Each call may spend up to the workspace's agent spend limit ($2.00 unless changed in Settings) without asking: it holds at most its estimate plus a margin while it runs and is charged what it bought, never more. A call whose maximum is above the limit is refused before anything is bought, with its estimate and maximum: ask the user, then call again with approvedMaxCostUsd set to the amount they approved.
Inputs: domain (required), projectId (required), scope, languageCode, +3 more. Returns: cost, meta, scope, domain, +4 more
get_google_ads_breakdownsRead Google Ads breakdownsARead-onlyInspect
Where and when the connected Google Ads account's ads ran in a period, from what the daily pull stored: the account's impressions, clicks, cost and conversions by device (MOBILE, DESKTOP, TABLET, CONNECTED_TV), by network (SEARCH, SEARCH_PARTNERS, CONTENT, YOUTUBE_SEARCH, YOUTUBE_WATCH, MIXED) and by day of the week, most spend first. Device and network numbers are stored from numbersStartOn; the day of the week covers every stored campaign day. Money is in the account's currency (currencyCode). Read-only; uses no credits.
Inputs: projectId (required), range, endDate, startDate. Returns: ok, meta, reason, devices, +14 more
get_google_ads_campaignsRead Google Ads campaignsARead-onlyInspect
The connected Google Ads account's campaigns for a period, from what the daily pull stored: impressions, clicks, click through rate, average cost per click, cost, conversions, conversion rate, cost per conversion and conversion value, most spend first, with each campaign's details as Google has them now (type, status, bid strategy and daily budget). Campaign numbers reach back 13 months. With withImpressionShare it also asks Google for each Search campaign's impression share and the share lost to budget and to rank (impressionShare says whether Google answered); nothing else here calls Google. Money is in the account's currency (currencyCode). Read-only; uses no credits.
Inputs: projectId (required), range, endDate, startDate, +1 more. Returns: ok, meta, rows, reason, +13 more
get_google_ads_conversion_actionsRead Google Ads conversionsARead-onlyInspect
The connected Google Ads account's conversions split by conversion action (purchase, sign-up, call and so on) for a period, from what the daily pull stored: conversions, conversion value and each action's share of all conversions, most conversions first. Read-only; uses no credits.
Inputs: projectId (required), range, endDate, startDate. Returns: ok, meta, rows, reason, +12 more
get_google_ads_landing_pagesRead Google Ads landing pagesARead-onlyInspect
The landing pages of the connected Google Ads account's ads that brought paid conversions in a period, from what the daily pull stored: clicks, cost, conversions and cost per conversion, most conversions first, with the page's organic clicks, impressions and average position from Search Console over the same days (null where Search Console has none for the page; organicState says whether it was read). Landing page numbers reach back about 90 days (numbersStartOn says from when). Money is in the account's currency (currencyCode). Read-only; uses no credits.
Inputs: projectId (required), limit, range, endDate, +1 more. Returns: ok, meta, rows, reason, +13 more
get_google_ads_search_termsRead Google Ads search termsARead-onlyInspect
The connected Google Ads account's paid search terms for a date range, from what the daily pull stored: impressions, clicks, cost, conversions, conversion value and cost per conversion, most spend first, with each term's organic clicks, impressions and average position from Search Console over the same days (null where Search Console has none for the term; organicState says whether it was read). Also the account's campaign spend (accountSpend) and, for Performance Max campaigns, Google's Performance Max search categories (searchCategories): groups of similar searches with impressions, clicks and conversions but no cost. Money is in the account's currency (currencyCode). Days are in the account's time zone; at most 90 days. Read-only; uses no credits.
Inputs: projectId (required), limit, endDate, startDate. Returns: ok, meta, note, rows, +14 more
get_google_ads_terms_without_rankingRead paid terms without a rankingARead-onlyInspect
The connected Google Ads account's paid search terms that the site has no organic position for, most spend first, with cost and conversions for a date range, and each term's organic clicks, impressions and Search Console average position where Search Console has them (organicState says whether it was read): the spend organic work could replace. Positions come from the project's rank tracking of its own site, read from saved checks; nothing is bought or fetched. tracked: true means the last check did not find the site; tracked: false means the term is not tracked, so its position is unknown (add it to rank tracking to check). Performance Max search categories are not included: they group searches and are not exact terms a rank check can look up. Money is in the account's currency (currencyCode); at most 90 days. Read-only; uses no credits.
Inputs: projectId (required), limit, endDate, startDate. Returns: ok, meta, note, rows, +14 more
get_google_analytics_audience_breakdownRead audience breakdownARead-onlyInspect
Sessions and engagement split by device, country, or new versus returning visitors, optionally compared with the previous period. Organic Search by default. Read only and free.
Inputs: projectId (required), limit, device, medium, +12 more. Returns: rows, error, quota, source, +12 more
get_google_analytics_ecommerce_performanceRead ecommerce performanceBRead-onlyInspect
Ecommerce results from Google Analytics by item or by landing page: transactions and revenue. Organic Search by default. Read only and free.
Inputs: projectId (required), limit, device, medium, +12 more. Returns: rows, error, quota, source, +12 more
get_google_analytics_key_eventsRead key eventsARead-onlyInspect
Key event counts from Google Analytics, by event or by event and landing page, optionally compared with the previous period. Organic Search by default. Read only and free.
Inputs: projectId (required), limit, device, medium, +12 more. Returns: rows, error, quota, source, +12 more
get_google_analytics_measurement_healthRead measurement healthARead-onlyInspect
Checks how the connected Google Analytics property is set up: web streams, key events, custom definitions and any configuration issues. Read only and free.
Inputs: projectId (required). Returns: error, issues, source, status, +5 more
get_google_analytics_organic_landing_pagesRead organic landing pagesARead-onlyInspect
Organic Search landing pages from Google Analytics with sessions, engagement, key events and ecommerce outcomes: use it for what organic visitors did on the site. get_landing_page_insights puts each page beside its Search Console clicks, and get_search_opportunities picks the pages closest to winning more clicks. Read only and free. Defaults to the last 28 complete property days.
Inputs: projectId (required), limit, device, medium, +9 more. Returns: rows, error, quota, source, +12 more
get_google_analytics_organic_overviewRead the organic overviewBRead-onlyInspect
Organic Search totals from Google Analytics for a date range next to the range just before it, with a daily or weekly trend. Read only and free.
Inputs: projectId (required), trend, device, country, +3 more. Returns: error, quota, trend, source, +8 more
get_google_analytics_page_performanceRead page performanceARead-onlyInspect
Page views, active users, engagement time and key events per page from Google Analytics. Organic Search by default; pass channel=all for every channel. Read only and free.
Inputs: projectId (required), limit, device, medium, +11 more. Returns: rows, error, quota, source, +12 more
get_google_analytics_site_searchRead site searchBRead-onlyInspect
What visitors type into the site's own search box, from Google Analytics, across all channels. Read only and free.
Inputs: projectId (required), limit, device, medium, +9 more. Returns: rows, error, quota, source, +12 more
get_google_analytics_traffic_acquisitionRead traffic acquisitionCRead-onlyInspect
Where sessions come from, by channel group, source and medium, or campaign, across all channels. Read only and free.
Inputs: projectId (required), limit, device, medium, +11 more. Returns: rows, error, quota, source, +12 more
get_google_analytics_traffic_breakdownRead traffic breakdownARead-onlyInspect
The Traffic Insights tabs from Google Analytics, with the same numbers as the page: channels (channel, source and medium, or campaign), ai_assistants (visits from AI assistants by source, landing page or key event), key_events, countries_devices (country, device, region or city) and pages. Every row comes with the period before it and the totals, and the text says when Google sampled, hid or folded numbers. Read only and free. Defaults to the last 28 complete property days.
Inputs: tab (required), projectId (required), group, device, +9 more. Returns: rows, error, group, state, +8 more
get_keyword_autocompleteGet autocomplete ideasAInspect
Buys Google Autocomplete for one keyword: the searches Google suggests as someone types it, in Google's order, long tail ideas in searchers' own words. Estimated price $0.01 per keyword, one request. Bought in the last 24 hours is read free, and every purchase opens free on Keyword Research's Autocomplete view; get_saved_keyword_autocomplete reads it without buying. Suggestions carry no volume: research_keywords or get_keyword_metrics price that. Each call may spend up to the workspace's agent spend limit ($2.00 unless changed in Settings) without asking: it holds at most its estimate plus a margin while it runs and is charged what it bought, never more. A call whose maximum is above the limit is refused before anything is bought, with its estimate and maximum: ask the user, then call again with approvedMaxCostUsd set to the amount they approved.
Inputs: keyword (required), projectId (required), languageCode, locationCode, +1 more. Returns: cost, meta, bought, keyword, +3 more
get_keyword_gapGet the keyword gapAInspect
Finds content gaps: the keywords a competitor ranks for in Google and your domain does not, highest search volume first, each with its search volume, difficulty and the competitor's position and page. Both sides are whole domains, subdomains included. One competitor a call; compare several with one call each. Estimated price: ~$0.05, one market data request for the top 100 keywords. The same gap and market saved within 12 hours is read free, and it opens free on Domain Overview's Content gap tab; get_saved_keyword_gap reads it without buying. Each call may spend up to the workspace's agent spend limit ($2.00 unless changed in Settings) without asking: it holds at most its estimate plus a margin while it runs and is charged what it bought, never more. A call whose maximum is above the limit is refused before anything is bought, with its estimate and maximum: ask the user, then call again with approvedMaxCostUsd set to the amount they approved.
Inputs: projectId (required), competitor (required), domain, languageCode, +2 more. Returns: cost, meta, saved, bought, +2 more
get_keyword_metricsGet keyword metricsAInspect
Metrics for a known list: returns search volume, keyword difficulty, intent, CPC, competition and monthly trends for up to 700 keywords you already have, in one call, with no new ideas. For ideas from a seed keyword, use research_keywords. For countries served from Google Ads data, difficulty and intent are null. Estimated price: ~$0.05 for 100 keywords and ~$0.20 for 700, doubled with clickstream, and ~$0.18 a call of up to 700 keywords for countries served from Google Ads data. The same keywords asked for again within 24 hours are answered from the saved copy at no charge. Bought stats also update keywords already in Saved Keywords. Each call may spend up to the workspace's agent spend limit ($2.00 unless changed in Settings) without asking: it holds at most its estimate plus a margin while it runs and is charged what it bought, never more. A call whose maximum is above the limit is refused before anything is bought, with its estimate and maximum: ask the user, then call again with approvedMaxCostUsd set to the amount they approved.
Inputs: keywords (required), projectId (required), sortBy, languageCode, +4 more. Returns: cost, meta, keywords, savedResultId
get_keyword_volume_historyGet keyword volume historyAInspect
Buys monthly Google search volume since 2021 for up to 100 keywords in one market: several years of demand, beyond the 12 months research shows. Estimated price: up to $0.03 for one keyword, up to $0.05 for 100, one request billed per keyword returned; keywords Google has no data for are free. The same list bought in the last 30 days is read free, and every purchase opens free on Keyword Research's Volume history view; get_saved_keyword_volume_history reads it without buying. Each call may spend up to the workspace's agent spend limit ($2.00 unless changed in Settings) without asking: it holds at most its estimate plus a margin while it runs and is charged what it bought, never more. A call whose maximum is above the limit is refused before anything is bought, with its estimate and maximum: ask the user, then call again with approvedMaxCostUsd set to the amount they approved.
Inputs: keywords (required), projectId (required), languageCode, locationCode, +1 more. Returns: cost, meta, bought, savedAt, +2 more
get_knowledge_baseRead the knowledge baseARead-onlyInspect
Reads what Visibility Genies keeps about the project's business and own website: the business memory by block (company, offers, audience, proof points, goals, positioning, voice, rules, notes and more), each entry with its short id, its state (Confirmed, or proposed by the AI, an assistant, Search Console, Analytics or a review) and its source, and the pages its site audit crawled, 50 at a time, each with its type, words and summary. Free. Read it before writing about the site.
Inputs: projectId (required), page, pageType. Returns: base, meta, pages, total, +1 more
get_landing_page_insightsRead landing page insightsARead-onlyInspect
Organic Traffic Insights: use it to see each landing page's search results and site results side by side. One row per landing page with Search Console clicks, impressions, CTR (0-1) and average position beside Google Analytics organic sessions, engaged sessions, engagement rate (0-1) and key events for the same period, plus the previous period's clicks and sessions. Pages match on host and path (scheme, a leading www., trailing slash, query string and fragment ignored). A null means that source has no row for the page, not zero; a side that is not connected is reported with a link and leaves the other side's rows in place. Rows are sorted by clicks, then sessions. For the queries behind a page, call get_search_console_performance with dimensions ['query'] and a page equals filter on each of its searchConsoleUrls. get_search_opportunities picks the pages closest to winning more clicks instead, and get_google_analytics_organic_landing_pages gives Analytics outcomes alone, with ecommerce. Read only and free.
Inputs: projectId (required), limit, device, offset, +5 more. Returns: ok, meta, rows, period, +8 more
get_link_metricsGet link metricsAInspect
Link metrics of up to 100 domains at once, as the Competitors and Link Prospects pages show them. Ranks and spam scores are 0 to 100 scores, not Google metrics. Estimated price: ~$0.15 for 10 competitors, ~$0.05 for 10 prospects, billed per request and per domain. The same domains saved within 6 hours are read free. Each call may spend up to the workspace's agent spend limit ($2.00 unless changed in Settings) without asking: it holds at most its estimate plus a margin while it runs and is charged what it bought, never more. A call whose maximum is above the limit is refused before anything is bought, with its estimate and maximum: ask the user, then call again with approvedMaxCostUsd set to the amount they approved.
Inputs: domains (required), projectId (required), set, approvedMaxCostUsd. Returns: set, cost, meta, rows, +1 more
get_local_rank_gridGet a local rank gridAInspect
Runs one Google Maps search for every point of a square grid centred on a coordinate and reports where the target business ranks at each point, with each point's result count and top listing. Use it to see how Maps visibility changes across an area. Looks 20 results deep per point. Estimated price: $0.04 for 3x3 (9 searches), $0.10 for 5x5 (25 searches). A grid bought with a cid opens on the Local page's Maps visibility tab. Each call may spend up to the workspace's agent spend limit ($2.00 unless changed in Settings) without asking: it holds at most its estimate plus a margin while it runs and is charged what it bought, never more. A call whose maximum is above the limit is refused before anything is bought, with its estimate and maximum: ask the user, then call again with approvedMaxCostUsd set to the amount they approved.
Inputs: center (required), target (required), keyword (required), projectId (required), +6 more. Returns: cost, grid, meta, summary, +2 more
get_local_serp_resultsGet local search resultsAInspect
Fetches a Google Maps or Local Finder results page from a coordinate, with the rows trimmed to identity, rank, rating, categories and hours. Decide yourself how to match a target business in the rows. Estimated price: $0.01 for a Google Maps search at any depth; Local Finder bills the same for each page of 20 places on desktop, 10 on mobile. Each call may spend up to the workspace's agent spend limit ($2.00 unless changed in Settings) without asking: it holds at most its estimate plus a margin while it runs and is charged what it bought, never more. A call whose maximum is above the limit is refused before anything is bought, with its estimate and maximum: ask the user, then call again with approvedMaxCostUsd set to the amount they approved.
Inputs: near (required), keyword (required), projectId (required), depth, +4 more. Returns: cost, meta, results, savedResultId
get_meta_ads_performanceRead Meta Ads performanceARead-onlyInspect
The connected Meta ad account's Facebook and Instagram ads for a date range, from what the daily pull stored: totals (spend, impressions, link clicks, CTR, CPC, CPM, purchases, leads, registrations, adds to cart and their value, ROAS), each campaign most spend first, and the landing pages the ads send people to. Reach shows for the last 30 or 90 stored days only, as Meta counts people per period. Money is in the account's currency (currencyCode). Results count within 7 days of a click or 1 day of a view. At most 90 days. Read-only; uses no credits.
Inputs: projectId (required), endDate, startDate. Returns: ok, meta, reason, totals, +11 more
get_page_viewRead a page's viewARead-onlyInspect
Pages: everything the project holds about one page of its own site, in one answer, to say how a page is doing. Returns the page facts (type, key page role, status code, indexable, words, last crawled, canonical), every tracked keyword whose latest check ranks this page with location, device, position and change since the check before, the tracked prompts whose AI answers cite it with engines, answers, whether the answer names the brand, last cited date and up to ten cited sentences, the open issues of the latest site audit on the page (muted and site level ones left out), and Search Console clicks, impressions, CTR (0-1 in the structured result) and position for the last 28 days against the 28 before with the top ten searches, plus Analytics organic sessions. Each part says when it has no data or is not connected. Read only and free.
Inputs: url (required), projectId (required), period. Returns: ok, meta, page, reason
get_project_contextRead project contextARead-onlyInspect
Reads the project's shared memory: the business memory by block (company, offers, audience, proof points, goals, positioning, voice, rules, notes and more), each entry with its short id, its state (Confirmed, or proposed by the AI, an assistant, Search Console, Analytics or a review) and its source; then competitors, key pages, brand names and the research log, and the blocks with nothing yet. Read it before planning work for a project. Free.
Inputs: projectId (required). Returns: meta, entries, keyPages, sections, +7 more
get_project_summaryRead the project summaryARead-onlyInspect
Where a project stands, in one call: the Summary page's six answers (site health, Google visits, what AI answers say, how strong your links are, who beats you, what to do this week), each with the newest day its data comes from, and Home's scores and changes with when each was measured. Read it first, before deciding what to buy. A project without data gets the same empty texts the pages show. Free: it reads saved data and the project's own Search Console connection, and buys nothing.
Inputs: projectId (required). Returns: meta, scores, changes, questions
get_ranked_keywordsGet ranked keywordsAInspect
Returns the keywords a domain or page ranks for in a country-level market: keyword, position, search volume, traffic, CPC, keyword difficulty and URL. By default it asks for Domain Overview's Top keywords page (the 100 keywords with the most traffic), so a page bought here opens free in the app and one bought in the app, or for a new rank tracker's suggestions, is answered here at no charge. Use get_domain_overview for a domain's totals. Estimated price: ~$0.05 for 100 rows, billed by the rows returned; asking again within 12 hours is answered from the saved copy at no charge. Each call may spend up to the workspace's agent spend limit ($2.00 unless changed in Settings) without asking: it holds at most its estimate plus a margin while it runs and is charged what it bought, never more. A call whose maximum is above the limit is refused before anything is bought, with its estimate and maximum: ask the user, then call again with approvedMaxCostUsd set to the amount they approved.
Inputs: target (required), projectId (required), limit, scope, +11 more. Returns: cost, meta, scope, target, +3 more
get_rank_historyRead rank historyARead-onlyInspect
What moved in a rank tracker between two days: for each keyword and device, its position on the first day and on the last (its newest check by the end of each day), and the change, positive when it moved up, as the Rank Tracking page counts it. Sorted by the size of the move. Give trackingKeywordId to add every check of one keyword between the two days. Positions reach back only as far as the plan keeps history. Free: it reads saved checks and buys nothing; run_rank_tracker buys new ones.
Inputs: projectId (required), trackerId (required), limit, device, +5 more. Returns: meta, rows, endDate, history, +3 more
get_rank_trackerRead rank trackersARead-onlyInspect
Read-only view of the project's rank trackers and their latest results. Uses no credits. When no tracker exists yet, call create_rank_tracker, then manage it with add_rank_tracking_keywords, remove_rank_tracking_keywords, estimate_rank_tracker_cost or run_rank_tracker. lastCheckedAt shows how fresh the positions are. With trackerId the result also carries each keyword's trackingKeywordId, which removals need. A position marked unconfirmed is a drop from a single reading that the next check has not yet confirmed.
Inputs: projectId (required), trackerId. Returns: meta, config, configs, results, +1 more
get_saved_ai_mentions_over_timeRead saved AI mentions over timeARead-onlyInspect
Reads the AI mentions over time already bought for a brand or domain in this project: its mentions and AI search volume in ChatGPT and Google's AI Overviews, month by month, in the United States. Uses no credits and never buys: a platform that is not saved or out of date is named with the price Brand Lookup's refresh asks for it, and get_ai_mentions_over_time buys it at that price, as the page does.
Inputs: query (required), projectId (required), scope. Returns: meta, toBuy, platforms, toBuyPriceUsd
get_saved_backlink_gapRead a saved backlink gapARead-onlyInspect
Reads a backlink gap saved by the Backlink Gap page or get_backlink_gap, however old, a page of prospects at a time, for free. It never buys: a competitor not saved for the domain is named with the price get_backlink_gap would pay for it.
Inputs: projectId (required), competitors (required), limit, domain, +1 more. Returns: meta, rows, saved, domain, +1 more
get_saved_brand_comparisonRead a saved brand comparisonARead-onlyInspect
Reads Brand Lookup's Competitor Research already bought for this project, however old, for free: a brand and its competitors side by side in AI answers (United States, English; ChatGPT and Google AI Overview), each one's mentions and AI search volume on each platform, and its Share of Voice. It never buys: what is not saved is named with the price compare_brands would pay for it.
Inputs: query (required), projectId (required), competitors (required), names, +1 more. Returns: meta, query, saved, comparison, +1 more
get_saved_brand_lookupRead a saved brand lookupARead-onlyInspect
Reads a Brand Lookup already bought for this project: how ChatGPT and Google's AI Overviews mention a brand or domain in the United States, and Share of Voice against the competitors it compared (get_saved_brand_comparison shows its Competitor Research). The pages they cite and sample prompts are bought on their own, so a lookup has them only when they were. Uses no credits and never buys: what is not saved or out of date is named with the price Brand Lookup asks for it, each of those two parts at its own. lookup_brand buys them at that price, as the page does, and compare_brands buys Competitor Research alone.
Inputs: query (required), projectId (required), names, scope, +1 more. Returns: meta, toBuy, lookup, addOnsToBuy, +1 more
get_saved_brand_mentionsRead saved brand mentionsARead-onlyInspect
Reads Brand Mentions already bought for a brand in this project, however old, for free: the web pages that mention it, their tone, the sites that mention it most and mentions by month. It never buys: what is missing or out of date is named with the price get_brand_mentions would pay for it.
Inputs: projectId (required), brand, excludeDomain. Returns: meta, saved, keyword, summary, +2 more
get_saved_brand_performanceRead saved brand performanceARead-onlyInspect
Reads Brand Performance already bought for a topic in this project, however old, for free: the brands and brand categories ChatGPT's answers about it mention most (United States, English; ChatGPT), and where a brand stands among them. It never buys: what is missing or out of date is named with the price get_brand_performance would pay for it.
Inputs: topic (required), projectId (required), brand. Returns: meta, saved, topic, brands, +1 more
get_saved_content_briefsRead content briefsARead-onlyInspect
Reads the briefs saved in Content Creation: each with its keyword, title, state and the Writing Assistant draft written from it; or one brief's title, meta description, reader, search intent, angle, what the pages that rank cover and miss, and its outline with each heading's terms, target length and whether its text is written. Uses no credits and calls no provider. No agent tool writes a brief or runs Content Creation: the page does, pricing each run first. save_draft can replace the text of the draft written from a brief.
Inputs: projectId (required), briefId. Returns: meta, brief, briefs
get_saved_content_templateRead a content templateARead-onlyInspect
Reads a keyword's content template, from its top 10 Google results: the typical length of the pages (the middle half of their word counts, a fact and not a target: Google sets no word count), the median number of subheadings and the median readability (0 hard to 100 easy), which help readers and are not Google ranking factors, the terms most of the pages use, each in all its forms, with how many use each, and each page with its headings, or why it could not be read. Length and terms are compared only when at least 5 pages could be read, as saved for this project, or lists every saved template when no keyword is given. Uses no credits and never buys: a template not saved is named with its price, which build_content_template or the Content Template page pays.
Inputs: projectId (required), keyword, languageCode, locationCode. Returns: meta, bought, savedAt, template, +1 more
get_saved_domain_comparisonRead a saved domain comparisonARead-onlyInspect
Reads a comparison of 2 to 5 domains saved by the Compare Domains page or compare_domains, however old, for free. It never buys: what is missing or out of date is named with the price compare_domains would pay for it.
Inputs: domains (required), projectId (required), languageCode, locationCode. Returns: meta, rows, saved, domains
get_saved_draftsRead draftsARead-onlyInspect
Reads the drafts saved in the Writing Assistant: each with its title, the content template it follows, its word count, its match with the top pages out of 100 against that template (the share of the terms most of the pages that rank use that it covers; length, headings and readability are not part of it), and when it was saved (updatedAt); or one draft's text, its match, what to do next, and the tone and originality notes saved for it. A draft names the brief Content Creation wrote it from, and when an agent last saved it with save_draft. Uses no credits and calls no provider. save_draft saves an article you wrote; no agent tool buys notes: the page does, pricing each run first.
Inputs: projectId (required), draftId. Returns: meta, draft, drafts
get_saved_keyword_autocompleteRead saved autocomplete ideasARead-onlyInspect
Reads Google Autocomplete suggestions already bought for a keyword in this project, however old, for free. It never buys: a keyword not saved is named with the price get_keyword_autocomplete would pay.
Inputs: keyword (required), projectId (required), languageCode, locationCode. Returns: meta, bought, keyword, savedAt, +1 more
get_saved_keyword_gapRead a saved keyword gapARead-onlyInspect
Reads a keyword gap saved by Domain Overview's Content gap tab or get_keyword_gap for this domain, competitor and market, for free. It never buys: a gap that is not saved, or no longer current, is named with the price get_keyword_gap would pay for it.
Inputs: projectId (required), competitor (required), domain, languageCode, +1 more. Returns: meta, saved, bought, keywords
get_saved_keyword_strategyRead a keyword strategyARead-onlyInspect
Reads a keyword strategy saved on the Keyword Strategy page: the project's keywords grouped into topics, each with its primary keyword, secondary keywords, total volume, average difficulty, intent, the page that should target it and its warnings (mixed search intents, a page two topics share, keywords whose saved SERP analyses put them on one page or apart), plus the keywords in no topic, with navigational searches apart. Also lists every saved strategy. Uses no credits and calls no provider. No agent tool builds a strategy: the grouping runs on the page, which prices it first.
Inputs: projectId (required), strategyId. Returns: meta, topics, strategy, ungrouped, +2 more
get_saved_keyword_volume_historyRead saved keyword volume historyARead-onlyInspect
Reads monthly Google search volume history already bought for exactly this list of keywords in this project, however old, for free. It never buys: a list not saved is named with the price get_keyword_volume_history would pay.
Inputs: keywords (required), projectId (required), languageCode, locationCode. Returns: meta, bought, savedAt, keywords
get_saved_page_checksRead saved page checksARead-onlyInspect
Reads the On Page SEO checks saved for this project, each ideas for one page of the project and the keyword it should rank for, from comparing it with the top 10 Google results: content (title, main heading, meta description, indexing), terms to cover (every term most top pages use that the page misses, in one row), internal links (the links to it from its own site, from Site Audit), backlinks (referring domains beside the top pages') and Core Web Vitals, each with the numbers behind it, ranked in that order of what matters for search, a missing meta description last since it does not affect ranking; then, for readers and not numbered with the ideas, since they are not Google ranking factors, its length, headings and readability beside the top pages', as facts; and the page's match with the top pages out of 100, the share of the keyword's content template terms it covers: one by its id, or a page's newest for a keyword, or else the list, newest first. Uses no credits and never buys: an On Page SEO check not saved is named with its price, which check_page_seo or the On Page SEO Checker page pays.
Inputs: projectId (required), checkId, keyword, pageUrl, +2 more. Returns: meta, check, bought, checks, +1 more
get_saved_perceptionRead a saved perception analysisARead-onlyInspect
Reads a perception analysis saved on the Perception page: how the AI answers the project already bought (Prompt Explorer answers, or one tracked prompt's runs) describe a brand and its competitors. Each brand's sentiment with a score from -100 to 100 and the answers behind it, the themes behind that with a quote found in one answer and the cited sites that drive each theme, and for a tracked prompt each brand's score run by run. Also lists every saved analysis. Uses no credits and calls no provider. No agent tool runs an analysis: it runs on the page, which prices it first.
Inputs: projectId (required), analysisId. Returns: meta, brands, themes, analyses, +2 more
get_saved_prompt_answersRead saved prompt answersARead-onlyInspect
Reads the answers Prompt Explorer already bought for one prompt in this project: what the ChatGPT, Claude, Gemini and Perplexity APIs (and Grok, DeepSeek and Muse Spark, when asked) answered, with the sources they cited, and what the ChatGPT and Gemini apps show a person. Uses no credits and never buys: an answer that is not saved is named with the price a Run on Prompt Explorer asks for it, and ask_ai_models buys it at that price, as a Run does. For the same prompt answered again over time, get_tracked_prompts reads Prompt Tracking's runs.
Inputs: prompt (required), projectId (required), apps, models, +3 more. Returns: meta, toBuy, answers, unreadable, +1 more
get_saved_questionsRead saved questionsARead-onlyInspect
Reads Questions already bought for this project: up to 200 questions people search about a topic (found by question words in the market's language: English, German, French, Spanish, Italian, Dutch and Portuguese have their own, and other languages use the English ones), most searched first, each with Google search volume, difficulty, AI search volume, and the People Also Ask question on the topic's saved SERP analysis it rewords (about 70% of its words shared), with each People Also Ask question listed. Uses no credits and never buys: questions or AI search volume not saved are named with their price, which find_questions or the Questions page pays.
Inputs: topic (required), projectId (required), languageCode, locationCode. Returns: meta, bought, savedAt, questions, +1 more
get_saved_topic_researchRead saved topic researchARead-onlyInspect
Reads a Topic Research already bought for this project: up to 300 keywords related to a topic, grouped into subtopics (by Google's product and service categories, or by the words they share), each with its keywords, total search volume, average difficulty, the questions among them, and the headlines that rank for its top keyword where a SERP analysis is saved. Uses no credits and never buys: a topic not saved is named with its price, which research_topic or the Topic Research page pays.
Inputs: topic (required), projectId (required), languageCode, locationCode. Returns: meta, bought, savedAt, research
get_saved_web_mentionsRead saved web mentionsARead-onlyInspect
Reads the web pages that name the project's brand, or their summary (view "summary"), as saved by Link Prospects' Mentions tab, Perception's "On the web" or get_web_mentions, however old, for free. It never buys: when nothing is saved it names the price get_web_mentions would pay.
Inputs: projectId (required), view, brand, limit, +1 more. Returns: meta, rows, view, brand, +2 more
get_scheduled_runsRead scheduled runsARead-onlyInspect
Everything in the workspace that spends on its own: scheduled rank checks, audits with Lighthouse, tracked prompts, competitor watches and AI share of voice refreshes, each with its project, how often it runs, its next run, the most a run and a month of runs may cost as its page quotes it, and whether it waits for approval. Schedules are approved and changed on their pages, never by an agent. Only a member with spend access can read it, within their allowed projects. Free: buys nothing.
Inputs: no named parameters. Returns: schedules, upToPerMonthUsd
get_search_console_performanceRead Search Console performanceARead-onlyInspect
Reads the project's Google Search Console search analytics: clicks, impressions, CTR and average position, grouped by query, page, country, device, date or search appearance, with filters, paging and a comparison with an earlier period. Needs Search Console connected to the project. Uses no credits.
Inputs: projectId (required), type, brand, compare, +8 more. Returns: ok, meta, rows, reason, +10 more
get_search_console_sitemapsRead Search Console sitemapsARead-onlyInspect
Lists the sitemaps submitted for the project's Search Console property: when each was sent and last read, how many addresses Google counted, and its errors and warnings. Submitting a sitemap is done at Google. Needs Search Console connected. Uses no credits.
Inputs: projectId (required). Returns: ok, meta, reason, siteUrl, +3 more
get_search_opportunitiesFind search opportunitiesARead-onlyInspect
Pages behind Search Console striking-distance queries (the best page for the query ranks 5 to 20, read from up to 25,000 query and page rows), joined with Google Analytics organic landing-page outcomes and scored by demand, business value and reachability: use it to pick the pages worth improving first. Counts are pages, except gscRowsConsidered. Read only and free.
Inputs: projectId (required), limit, endDate, startDate. Returns: rows, error, quota, source, +9 more
get_serp_historyGet search results historyAInspect
Google's results for a keyword as the data service saved them during one past month: each snapshot's date, the SERP features on the page and its top 20 organic results with position, domain, URL and title. Shows who ranked where before, to see whether rankings moved. The data service keeps the last 12 months. Estimated price: up to $0.01, one request for a month's snapshots. A month saved within 30 days is read free, and it opens free in Keyword Research's SERP view. Each call may spend up to the workspace's agent spend limit ($2.00 unless changed in Settings) without asking: it holds at most its estimate plus a margin while it runs and is charged what it bought, never more. A call whose maximum is above the limit is refused before anything is bought, with its estimate and maximum: ask the user, then call again with approvedMaxCostUsd set to the amount they approved.
Inputs: month (required), keyword (required), projectId (required), languageCode, +2 more. Returns: cost, meta, bought, snapshots, +1 more
get_serp_resultsGet search resultsAInspect
Fetch Google's organic search results for 1-10 keywords, or with type their Google News articles or Google Images results. Use this to inspect who ranks for a query, verify competitors, compare SERPs across keywords, or gather source URLs before content planning. Returns the top depth organic rows per keyword (default 20). Estimated price per keyword: $0.01 at the default depth of 20, $0.04 at 100; News $0.01 for 20 articles, Images $0.01. Results are the SERP analysis Keyword Research shows: a keyword bought here opens free there, and a keyword it holds a current copy of, as deep as asked, is answered here at no charge for 12 hours. Per-keyword errors don't fail the batch. Each call may spend up to the workspace's agent spend limit ($2.00 unless changed in Settings) without asking: it holds at most its estimate plus a margin while it runs and is charged what it bought, never more. A call whose maximum is above the limit is refused before anything is bought, with its estimate and maximum: ask the user, then call again with approvedMaxCostUsd set to the amount they approved.
Inputs: queries (required), projectId (required), type, depth, +1 more. Returns: cost, meta, results, savedResultId
get_spend_summaryRead spendBRead-onlyInspect
The account's balance (free to spend, or sponsored when the account manager pays) and what it was charged in a month, this month so far unless one is given, from the same wallet the Spend page reads: by feature, by project, by client and by who started the work (MCP agents, pages in the app, scheduled runs), with what is still held for work running now and, on the account's own keys, what those keys paid. For the Owner and people who manage billing, also each client's month at the client price the account set for it (its own pricing toward that client, which changes nothing it is charged), and what its budget has left when one is on. Also the agent spend limit: how much one tool call may spend without asking. Only a member with spend access can read it, within their allowed projects. Free: buys nothing.
Inputs: month. Returns: meta, month, balance, heldUsd, +7 more
get_subdomainsGet subdomainsAInspect
Lists a domain's subdomains in one market, most organic traffic first: each subdomain's estimated organic and paid traffic, its organic and paid keyword counts and its keywords in the top 3 and top 10. Estimated price: up to $0.05, one market data request for up to 100 subdomains, billed by the rows returned. The same domain and market saved within 12 hours is read free, and it opens free on Domain Overview's Subdomains tab. Each call may spend up to the workspace's agent spend limit ($2.00 unless changed in Settings) without asking: it holds at most its estimate plus a margin while it runs and is charged what it bought, never more. A call whose maximum is above the limit is refused before anything is bought, with its estimate and maximum: ask the user, then call again with approvedMaxCostUsd set to the amount they approved.
Inputs: projectId (required), domain, languageCode, locationCode, +1 more. Returns: cost, meta, bought, savedResultId
get_tiktok_ads_performanceRead TikTok Ads performanceARead-onlyInspect
The connected TikTok advertiser's results for a date range, from what the daily pull stored: totals and per-campaign spend, impressions, clicks, CTR, CPC, CPM, conversions, cost per conversion, purchases, purchase value and ROAS, most spend first. Money is in the advertiser's currency (currencyCode); conversions follow the ad account's attribution setting. Days are in the advertiser's time zone; at most 90 days. Read-only; uses no credits.
Inputs: projectId (required), endDate, startDate. Returns: ok, meta, reason, totals, +10 more
get_tracked_promptsRead tracked promptsARead-onlyInspect
Reads Prompt Tracking in this project: the prompts asked again on a schedule, the brand each watches for, and, for one tracked prompt, every run with how many of each model's or app's answers over a period name the brand (the visibility rate, with the number of answers), each answer's rank for the brand in the list that names it ('2nd of 5 in its list', items counted whatever they name), which varies from run to run, and the sites the answers cite most, the project's own marked. Uses no credits and calls no provider. No agent tool sets up, runs or changes a tracked prompt: the Prompt Tracking page prices each schedule and the person approves it there.
Inputs: projectId (required), period, trackedPromptId. Returns: meta, trackedPrompt, trackedPrompts
get_traffic_historyGet traffic historyAInspect
Estimated organic and paid search traffic of up to 10 domains, month by month over the last 12 months, in one market. Estimated price: ~$0.25 for 2 domains, ~$0.27 for 10: one request, billed by the domains asked for. The same domains and market saved within a day are read free, and it opens free on Compare Domains. Each call may spend up to the workspace's agent spend limit ($2.00 unless changed in Settings) without asking: it holds at most its estimate plus a margin while it runs and is charged what it bought, never more. A call whose maximum is above the limit is refused before anything is bought, with its estimate and maximum: ask the user, then call again with approvedMaxCostUsd set to the amount they approved.
Inputs: domains (required), projectId (required), languageCode, locationCode, +1 more. Returns: cost, meta, bought, savedResultId
get_tripadvisor_reviewsGet Tripadvisor reviewsAInspect
Collects Tripadvisor reviews for one listing, with rating, author, visit date, text, and whether the owner responded. Use it for review-gap analysis against competitors and to spot unanswered reviews; find the listing's url_path with search_tripadvisor_businesses first. Usually completes within this call; if the queued job is still running you get status 'processing' plus a taskId, so call again with that taskId in 30-60 seconds to collect the result at no extra cost. Estimated price: $0.01 for 20 reviews, billed per 10 at twice the Local page's normal-priority rate, so the call usually returns them; reviews bought within a day, here or on the Local page, are answered at no charge. Each call may spend up to the workspace's agent spend limit ($2.00 unless changed in Settings) without asking: it holds at most its estimate plus a margin while it runs and is charged what it bought, never more. A call whose maximum is above the limit is refused before anything is bought, with its estimate and maximum: ask the user, then call again with approvedMaxCostUsd set to the amount they approved.
Inputs: projectId (required), depth, sortBy, taskId, +2 more. Returns: cost, meta, status, taskId, +3 more
get_trustpilot_reviewsGet Trustpilot reviewsAInspect
Collects Trustpilot reviews for one company, with rating, author, verified status, text, and whether the company replied. Use it for review-gap analysis against competitors and to spot unanswered reviews; find the company's Trustpilot domain with search_trustpilot_businesses first. Usually completes within this call; if the queued job is still running you get status 'processing' plus a taskId, so call again with that taskId in 30-60 seconds to collect the result at no extra cost. Estimated price: $0.01 for 20 reviews, billed per 20 at twice the Local page's normal-priority rate, so the call usually returns them; reviews bought within a day, here or on the Local page, are answered at no charge. Each call may spend up to the workspace's agent spend limit ($2.00 unless changed in Settings) without asking: it holds at most its estimate plus a margin while it runs and is charged what it bought, never more. A call whose maximum is above the limit is refused before anything is bought, with its estimate and maximum: ask the user, then call again with approvedMaxCostUsd set to the amount they approved.
Inputs: projectId (required), depth, domain, sortBy, +2 more. Returns: cost, meta, status, taskId, +3 more
get_web_mentionsGet web mentionsAInspect
Finds web pages that name the project's brand (an exact phrase), for link building: each page's site, title, publish date, how it leans (positive, negative or neutral) and the site's 0 to 100 link-graph rank, strongest sites first and one page per site, with the project's own site left out. It cannot tell whether a page links to the project: a page is a prospect to check, and "site links to you" says yes only when the project's saved referring domains list the site. view "summary" instead returns how many pages name the brand, how they lean, and the 10 sites that name it most. Estimated price: up to $0.06 for pages (one request, billed by the up to 100 pages returned) or up to $0.05 for the summary. An answer saved within 7 days is read free; get_saved_web_mentions reads it without buying. save_link_prospect keeps a site as a prospect. Each call may spend up to the workspace's agent spend limit ($2.00 unless changed in Settings) without asking: it holds at most its estimate plus a margin while it runs and is charged what it bought, never more. A call whose maximum is above the limit is refused before anything is bought, with its estimate and maximum: ask the user, then call again with approvedMaxCostUsd set to the amount they approved.
Inputs: projectId (required), view, brand, limit, +2 more. Returns: cost, meta, rows, view, +4 more
get_whoisGet a Whois recordAInspect
Reads a domain's Whois record: when it was registered and changed, when it expires, its registrar and status codes, when Visibility Genies first saw it, and its referring domains and backlinks. Estimated price: ~$0.25, one Whois lookup. A record saved within 30 days is read free, and it opens free on Domain Overview. Each call may spend up to the workspace's agent spend limit ($2.00 unless changed in Settings) without asking: it holds at most its estimate plus a margin while it runs and is charged what it bought, never more. A call whose maximum is above the limit is refused before anything is bought, with its estimate and maximum: ask the user, then call again with approvedMaxCostUsd set to the amount they approved.
Inputs: projectId (required), domain, approvedMaxCostUsd. Returns: cost, meta, bought, savedResultId
inspect_urlsInspect URLsARead-onlyInspect
Asks Google's URL Inspection for up to 10 URLs of the project's Search Console property: whether each is indexed, its coverage state, and the canonical Google chose. A URL that fails is reported next to the ones that worked. Needs Search Console connected. Uses no credits.
Inputs: urls (required), projectId (required), languageCode. Returns: ok, meta, reason, results, +3 more
list_actionsList actionsARead-onlyDestructiveInspect
Read saved project actions in the order the Actions page shows them, with their practical steps, pages, due day, evidence date, assignment and status. Uses no credits or provider calls. Suggested work comes from saved audit, prompt, Perception, Search Console and gap evidence. Optional filters by status, source, subject, assignee and due day, and pagination, do not change the saved work. get_action reads one action with its notes and change history.
Inputs: projectId (required), dueTo, limit, offset, +5 more. Returns: meta, rows, assignees, totalCount
list_archived_projectsList archived projectsARead-onlyInspect
Lists archived projects visible to this connection, with their id, name, domain and organizationId. Free: buys no data. Pass an id as archivedProjectId to restore_project. A project with deleting=true is being permanently deleted and cannot be restored. Does not include active projects; use list_projects for those.
Inputs: no named parameters. Returns: projects
list_auditsList site auditsARead-onlyInspect
Lists a project's site audits, newest first, with each audit's ID, start URL, status, pages crawled, crawler identity and dates. Pass an ID to get_audit_status, get_audit_issues or get_audit_pages to read an earlier audit. Free: reads saved audit data. Audits older than the history the plan keeps are left out.
Inputs: projectId (required), limit. Returns: meta, total, audits
list_business_categoriesList business categoriesARead-onlyInspect
Lists the Google Business Profile category slugs (for example pizza_restaurant) with how many businesses use each, optionally narrowed to those containing some text. Uses no credits; the full list is kept for 7 days.
Inputs: projectId (required), limit, query. Returns: meta, categories
list_clientsList clientsARead-onlyInspect
Lists the clients the user's workspace groups its sites (projects) under: each client's id and name, and its sites by name and id. Free: buys no data. Pass a client's id to create_project as clientId to add a site under it, or to set_project_client to move a site into it. When the user belongs to several organizations, each client is labeled with its organization.
Inputs: no named parameters. Returns: meta, clients
list_link_prospectsList link prospectsARead-onlyInspect
Lists the project's link prospects, the sites it wants a link from, newest first: each with the competitors it links to, its status (to_contact, contacted, won, lost), notes, and its rank passed (the rank its links pass to the competitors, a 0 to 100 link-graph score, not a Google metric, which sitewide links inflate), average link spam and backlinks as its table showed them when it was added or filled in, with how many stand at each status. A prospect whose site, or a site it is part of, was marked for disavow in Backlinks after it was added is flagged disavowed, and one the project's saved referring domains list as already linking to it is flagged links to you: each is kept until the user removes it. Uses no credits and calls no provider.
Inputs: projectId (required), limit, offset, status. Returns: meta, rows, counts, totalCount
list_projectsList projectsARead-onlyInspect
Lists the projects you can work in. It buys no data and uses no credits. Call it whenever another Visibility Genies tool needs a projectId: use the id of the project you want. Each project has {id, name, domain, locationCode, languageCode, clientId, clientName}. locationCode and languageCode are the project's default market, which tools use when a call leaves location and language out. clientId and clientName give the client the project is grouped under (see list_clients), or null. If you belong to several organizations, each project also names its organization and organizationId (pass that to create_project). Projects that other accounts shared with you are listed as well, each with shared {ownerAccount, level}: the level (view, work, work_and_spend) is what you may do there, and its paid tools are charged to the owner's account, within the owner's agent spend limit.
Inputs: no named parameters. Returns: meta, projects
list_saved_keywordsList saved keywordsARead-onlyInspect
Lists the keywords saved to a project with the volume, difficulty and CPC saved for each, and their tags, newest first. Reads the database only: no data is bought and no credits are used.
Inputs: projectId (required), tags, limit, search. Returns: meta, rows, tags, totalCount
list_saved_prompt_answersList saved prompt answersARead-onlyInspect
Lists every answer Prompt Explorer bought for this project, a run at a time, newest first: the prompt, how it was asked, and each model's or app's answer with its sources. Runs of tracked prompts are listed too, marked as Prompt Tracking samples (source: tracking). Answers stay saved for good, so what a model said months ago is read here. Uses no credits.
Inputs: projectId (required), limit, offset. Returns: meta, runs, hasMore
list_saved_resultsList saved resultsARead-onlyInspect
Lists the answers this project's paid tool calls returned, newest first: when, which tool, what was asked, what it cost and the first sentence of its answer. Every answer a tool that buys data returns is saved, except those of run_rank_tracker and run_site_audit, whose work is saved with the tracker or audit. Read one again with read_saved_result instead of buying it again. Uses no credits.
Inputs: projectId (required), tool, limit, offset. Returns: meta, hasMore, results
lookup_brandLook up a brandAInspect
Buys a Brand Lookup for this project, as the Brand Lookup page does (United States, English; ChatGPT and Google AI Overview): how often AI answers mention a brand or domain and its AI search volume on each platform, and, when asked, the pages those answers cite and sample prompts, with Competitor Research against any competitors named. Estimated price: ~$0.44 for the mentions and AI search volume; cited pages ~$0.44 and sample prompts ~$0.80 more; Competitor Research ~$0.44 with 1 competitor, ~$0.44 with 5. What is saved and current, here or on the page, is read free; only the parts missing or out of date (after 24 hours) are bought, as the page's Refresh buys them. It opens free on Brand Lookup and in its History; get_saved_brand_lookup reads it without buying. Each call may spend up to the workspace's agent spend limit ($2.00 unless changed in Settings) without asking: it holds at most its estimate plus a margin while it runs and is charged what it bought, never more. A call whose maximum is above the limit is refused before anything is bought, with its estimate and maximum: ask the user, then call again with approvedMaxCostUsd set to the amount they approved.
Inputs: query (required), projectId (required), names, scope, +4 more. Returns: cost, meta, query, bought, +4 more
move_actionMove an actionAInspect
Move a saved action up or down within its own status column on the Actions board, such as To do. To move it to another column, use update_action with a status. Uses no credits or provider calls.
Inputs: id (required), position (required), projectId (required). Returns: meta, action
quote_competitor_refreshQuote a competitor refreshAInspect
Create a saved quote without provider calls. Returns exact components, limits, payer, setup exclusions, expiry and confirmation hash. Absent competitorIds uses Tracking; explicit [] stays empty. promptIds are required and [] buys no AI answers. Watch quotes cover full future refreshes even if caches are fresh; the Owner approves a watch on the Competitors page. Show the entire quote and paying account to the user before confirmation.
Inputs: projectId (required), promptIds (required), kind, market, +1 more. Returns: meta, result
read_competitor_refreshRead competitor refresh progressBRead-onlyInspect
Read saved component outcomes, actual spend, payer and safe continuation state. Does not execute or buy anything.
Inputs: id (required), projectId (required). Returns: meta, result
read_competitorsRead competitorsBRead-onlyInspect
Read Tracking, Suggested or Ignored competitors, saved Google and AI measurements, evidence, gaps, setup and stable shared defaults. Optional competitorId opens head-to-head comparison. Uses no credits or provider calls. Missing measurements stay null, never zero.
Inputs: projectId (required), cursor, market, status, +2 more. Returns: meta, result
read_saved_resultRead a saved resultARead-onlyInspect
Reads a saved result of a paid tool call again, a page at a time: a result cut short to fit your context is read to the end here, never bought again. An answer that fits one page comes back whole; a bigger one a page of its largest list at a time, with the rest of the answer beside it. Uses no credits.
Inputs: resultId (required), projectId (required), list, limit, +1 more. Returns: list, meta, rows, lists, +5 more
recover_competitor_refreshRecover a competitor refreshAInspect
Reconcile interrupted work from saved evidence and ledger records only. No provider requests or new purchases. Uncertain outcomes must remain uncertain until proven; never automatically rebuy them.
Inputs: id (required), projectId (required). Returns: meta, result
remove_competitorRemove a competitorADestructiveInspect
Remove a saved competitor after the user requests it. Use status ignored via save_competitor to suppress a suggestion instead. Uses no credits.
Inputs: id (required), projectId (required), expectedRevision. Returns: meta, result
remove_rank_tracking_keywordsRemove rank tracking keywordsADestructiveInspect
Stops tracking keywords on a rank tracker while keeping their position history. IDs that are unknown, stale, from another tracker or repeated are skipped, so removed may be smaller than requested. Uses no credits.
Inputs: projectId (required), trackerId (required), keywordIds (required). Returns: meta, removed, requested, trackerId, +1 more
research_keywordsResearch keywordsAInspect
Ideas from seeds: finds related keyword ideas with search volume, difficulty and CPC for 1-5 seed keywords in one call. For metrics on keywords you already have, use get_keyword_metrics. Estimated price per seed: ~$0.06 to $0.18 at 150 results, ~$0.10 to $0.29 at 300, ~$0.15 to $0.44 at 500 (Auto buys one to three sources), doubled with clickstream; ~$0.18 at any limit for countries served from Google Ads data, where difficulty/intent are unavailable. Asking again within 24 hours is answered from the saved copy at no charge: in Google Ads countries at any result limit, elsewhere for the same result limit and clickstream setting. Returns per-seed results, each with a link that reopens it in Keyword Research; a single bad seed won't fail the batch. Each call may spend up to the workspace's agent spend limit ($2.00 unless changed in Settings) without asking: it holds at most its estimate plus a margin while it runs and is charged what it bought, never more. A call whose maximum is above the limit is refused before anything is bought, with its estimate and maximum: ask the user, then call again with approvedMaxCostUsd set to the amount they approved.
Inputs: seeds (required), projectId (required), resultLimit, approvedMaxCostUsd, +1 more. Returns: cost, meta, results, savedResultId
research_topicResearch a topicAInspect
Buys Topic Research: up to 300 keywords related to a topic, grouped into subtopics (by Google's product and service categories, or by the words they share), each with its keywords, total search volume, average difficulty, the questions among them, and the headlines that rank for its top keyword where a SERP analysis is saved. Estimated price up to $0.10 per topic. A topic researched in the last 24 hours is answered from its saved copy at no charge, and every purchase opens free on the Topic Research page. Headlines are only read, never bought. Each call may spend up to the workspace's agent spend limit ($2.00 unless changed in Settings) without asking: it holds at most its estimate plus a margin while it runs and is charged what it bought, never more. A call whose maximum is above the limit is refused before anything is bought, with its estimate and maximum: ask the user, then call again with approvedMaxCostUsd set to the amount they approved.
Inputs: topic (required), projectId (required), languageCode, locationCode, +1 more. Returns: cost, meta, bought, savedAt, +2 more
restore_actionRestore an actionAInspect
Put a dismissed action back on the Actions board, in To do or Suggested. Uses no credits or provider calls.
Inputs: id (required), projectId (required), status. Returns: meta, action
restore_projectRestore a projectAInspect
Restores an archived project and its saved data to the active project list. The call itself is free and buys no data; previously enabled schedules may resume and spend under their existing approvals. Pass archivedProjectId from list_archived_projects. Requires project deletion rights, as archiving does. A project whose permanent deletion has started cannot be restored.
Inputs: archivedProjectId (required). Returns: success, projectId
run_rank_trackerRun a rank checkAInspect
Starts one live rank check for every keyword and device on a tracker. Estimated price: about $0.20 per 10 keyword and device checks at the default depth with AI Overviews; estimate_rank_tracker_cost gives the tracker's total for free. That total with its margin is checked against the agent spend limit before anything starts; the checks are then charged as they run, never above the amount allowed. A check already running is reported and not charged, and the tracker's schedule is unchanged. Each call may spend up to the workspace's agent spend limit ($2.00 unless changed in Settings) without asking: it holds at most its estimate plus a margin while it runs and is charged what it bought, never more. A call whose maximum is above the limit is refused before anything is bought, with its estimate and maximum: ask the user, then call again with approvedMaxCostUsd set to the amount they approved.
Inputs: projectId (required), trackerId (required), approvedMaxCostUsd. Returns: cost, meta, runId, started, +3 more
run_site_auditRun a site auditAInspect
Starts a site audit. It crawls the site (same origin, following robots.txt), checks each page for SEO problems such as broken links, missing or duplicate titles and descriptions, redirect chains, orphan pages, canonical mistakes and pages blocked by robots.txt, and can run Lighthouse on a sample of pages. The crawl is free. Lighthouse is paid (up to 10 pages on mobile and desktop, up to 20 tests, up to $0.20) and is billed while the audit runs. The audit works in the background: poll get_audit_status, then read get_audit_issues, and get_audit_pages for Lighthouse scores. When a site rate limits the crawler it backs off and retries, and pages it still cannot read are marked blocked or rate-limited instead of being guessed at. Every audit crawls at one gentle pace, about a page a second, and crawlSpeed is accepted but ignored. The crawler slows down by itself; if the site keeps refusing, allow our crawler by name and address (the bot page lists both). The crawler calls itself VisibilityGeniesBot unless crawlerIdentity says otherwise: desktop or mobile send a browser user agent that keeps that token, while chrome_windows, firefox_windows, safari_macos, chrome_android and safari_iphone send an ordinary browser's headers without it, and none imitates a search engine. Pages are read as HTML, with no JavaScript, unless renderJavaScript is true. The project Owner may let audits of the project's own site ignore robots.txt, and those audits report a robots-txt-ignored finding. Each call may spend up to the workspace's agent spend limit ($2.00 unless changed in Settings) without asking: it holds at most its estimate plus a margin while it runs and is charged what it bought, never more. A call whose maximum is above the limit is refused before anything is bought, with its estimate and maximum: ask the user, then call again with approvedMaxCostUsd set to the amount they approved.
Inputs: url (required), projectId (required), maxPages, crawlMode, +7 more. Returns: cost, meta, auditId, savedResultId
save_competitorSave a competitorAInspect
Add or edit a competitor, aliases, notes and status. New manual rivals default to Tracking. Maximum ten tracked rivals for new additions; existing longer lists remain intact. Confirm a name-only suggestion's real domain before saving with status tracking. Explicit comparison selections always override these defaults. Use expectedRevision from read_competitors for edits.
Inputs: projectId (required), id, name, notes, +5 more. Returns: meta, result
save_competitorsSave several competitorsAInspect
Add up to 50 competitors in one call, each a website, a name, or both, with an optional note, as the Add several box does. Websites are normalized, the project's own site is left out, and a website already saved is updated in place. Rivals with a website are tracked while Tracking has room for ten; the rest are saved under Suggested. Returns how many are tracking, suggested and updated. Confirm the list with the user first. Uses no credits.
Inputs: projectId (required), competitors (required). Returns: meta, result
save_draftSave a draftADestructiveInspect
Saves an article you wrote in the project's Writing Assistant, where the user reads and edits it: a new draft, with a title and optionally a content template to score it against, or an update of a draft you read with get_saved_drafts. An update passes the draft's updatedAt as you read it and is refused if the draft was saved since, so a person's edits are never written over: read it again and save your changes onto that text. get_saved_drafts lists it as saved by an agent, also after a person edits it. Returns the draft's id, its link in the app, its new updatedAt, and its match with the top pages out of 100 against its template with what to do next, from the Writing Assistant's own measure. Uses no credits and calls no provider. A template from another project is refused.
Inputs: text (required), projectId (required), title, draftId, +2 more. Returns: meta, draft, created
save_keywordsSave keywordsADestructiveInspect
Saves up to 500 keywords to a project's saved list, and can tag them. Buys no data and uses no credits. Pass metrics to keep the numbers you already researched with them.
Inputs: keywords (required), projectId (required), tags, metrics, +3 more. Returns: meta, tags, tagMode, keywords, +5 more
save_link_prospectSave a link prospectADestructiveInspect
Adds a site to the project's link prospects, or changes a kept one's status or note. Uses no credits and calls no provider. A site already kept is never added twice: only the competitors it newly links to are added, and its status and note change only when given; a note given replaces the one kept. A site this tool adds is recorded as added by hand, without rank passed, spam score or backlinks. Adding it on the Backlink Gap page (or from a competitor's referring domains on the Backlinks page) keeps them, and fills them in for a site already kept without them, so point the user there when those metrics matter. The project's own domain and sites marked for disavow in Backlinks are refused.
Inputs: domain (required), projectId (required), note, status, +1 more. Returns: meta, added, prospect
search_knowledge_baseSearch the knowledge baseARead-onlyInspect
Finds the project's own pages that best match a keyword, topic or title, from the text its site audit crawled, with a summary and an excerpt of each. Free. Use it to see whether the site already covers a topic and which page does.
Inputs: query (required), projectId (required), limit. Returns: meta, pages
search_local_businessesSearch local businessesAInspect
Searches Business Listings near a coordinate, with optional rating, review and claim filters. It is not a Maps rank check: use get_local_serp_results for that. Estimated price: $0.04 for 20 businesses, billed per business returned. Each call may spend up to the workspace's agent spend limit ($2.00 unless changed in Settings) without asking: it holds at most its estimate plus a margin while it runs and is charged what it bought, never more. A call whose maximum is above the limit is refused before anything is bought, with its estimate and maximum: ask the user, then call again with approvedMaxCostUsd set to the amount they approved.
Inputs: near (required), projectId (required), limit, query, +7 more. Returns: cost, meta, businesses, savedResultId
search_tripadvisor_businessesSearch Tripadvisor businessesAInspect
Finds hotels, restaurants, attractions, and other businesses on Tripadvisor by name or category within a city or region, with rating, review count, category, and the url_path that get_tripadvisor_reviews takes. Usually completes within this call; a 'processing' response returns a taskId to call back with in 30-60 seconds at no extra cost. Estimated price: $0.01 for 30 results, billed per 30 at twice the Local page's normal-priority rate, so the call usually returns them. Each call may spend up to the workspace's agent spend limit ($2.00 unless changed in Settings) without asking: it holds at most its estimate plus a margin while it runs and is charged what it bought, never more. A call whose maximum is above the limit is refused before anything is bought, with its estimate and maximum: ask the user, then call again with approvedMaxCostUsd set to the amount they approved.
Inputs: projectId (required), depth, taskId, keyword, +3 more. Returns: cost, meta, status, taskId, +2 more
search_trustpilot_businessesSearch Trustpilot businessesAInspect
Finds companies on Trustpilot by name or category, with rating, review count, and the domain that get_trustpilot_reviews takes. Usually completes within this call; a 'processing' response returns a taskId to call back with in 30-60 seconds at no extra cost. Estimated price: $0.01 for 10 results, billed per 10 at twice the Local page's normal-priority rate, so the call usually returns them. Each call may spend up to the workspace's agent spend limit ($2.00 unless changed in Settings) without asking: it holds at most its estimate plus a margin while it runs and is charged what it bought, never more. A call whose maximum is above the limit is refused before anything is bought, with its estimate and maximum: ask the user, then call again with approvedMaxCostUsd set to the amount they approved.
Inputs: projectId (required), depth, taskId, keyword, +1 more. Returns: cost, meta, status, taskId, +2 more
set_project_clientSet a project's clientAInspect
Puts a site (project) under one of the workspace's clients, under a new client by name, or out of its client. Free: buys no data.
Inputs: clientId (required), projectId (required), newClientName. Returns: meta, project
update_actionUpdate an actionAInspect
Change a saved action's status, priority or assignee in this project. Set assigneeUserId to null to unassign, or use a current organization member id returned by list_actions. Marking done stamps its completion time; leaving done clears it. Dismissed audit findings return only above their dismissal baseline; other sources stay dismissed until manually restored. Uses no credits or provider calls.
Inputs: id (required), projectId (required), status, priority, +1 more. Returns: meta, action
update_knowledge_baseUpdate the knowledge baseADestructiveInspect
Proposes facts for the project's business memory and writes page summaries. Every AI feature in Visibility Genies gets the whole confirmed business memory, then the proposals marked as unconfirmed, and the pages that match its task. Free. Your proposals are used for suggestions at once, labelled as yours, and reach drafts and public text after the customer confirms them on the knowledge base page. A field given replaces the assistants' earlier open fact proposals in its block and never changes what the customer confirmed or an assistant's section proposals. To add one fact without replacing others, use proposeFact in update_project_context. Facts are short sentences on one line (500 characters at most, 20 a field), each optionally naming the page it came from. Summaries are at most 1,000 characters. Write only what the site states, and confirm with the user when unsure. Do not store figures read from Search Console, Analytics or Google Ads here (no clicks, impressions, positions, sessions, conversions or spend).
Inputs: projectId (required), fields, summaries. Returns: meta, unknownUrls
update_project_contextUpdate project contextADestructiveInspect
Changes the project's shared memory with a batch of edits, applied in order except competitor additions, which are saved after all the other edits: propose a business fact in a block (proposeFact: one line, with the page it came from as sourceUrl), propose a section or a custom section, withdraw an assistant's open proposal of one, add or remove competitors and key pages, set brand names, append or remove research log entries. Your proposals are used for suggestions at once, labelled as yours, and reach drafts and public text after the customer confirms them on the knowledge base page. Nothing you write changes what the customer confirmed. Do not store figures read from Search Console, Analytics or Google Ads here (no clicks, impressions, positions, sessions, conversions or spend); page topics and business facts are fine. Competitors added here are saved as the Add several box saves them: tracked while Tracking has room under the plan's limit, the rest suggested. Free.
Inputs: updates (required), projectId (required). Returns: meta, entries, keyPages, sections, +7 more
whoamiShow my accountARead-onlyInspect
Shows which account this connection acts as: the email, the token scopes, the account's balance (what is free to spend, or that the account is sponsored and its manager pays), on a paid plan that includes data, the data included in the plan that is left this month, and the agent spend limit per request. Free: buys no data.
Inputs: no named parameters. Returns: meta, mode, scopes, balance, +3 more
Tool Schema Changelog
Recent tool additions, removals, and schema changes observed during successful MCP inspections.
Loading tool schema changelog…
Related MCP Connectors
Content, rankings, AI citations and compliance checks for your site, from your AI assistant.
Run your company blog for SEO and AI search from your AI assistant.
Check your site's SEO, rankings and AI search visibility, and review fixes, from your AI assistant.
AI SEO agent on your Google Search Console data: audits, quick wins, content, WordPress publishing.
Related MCP Servers
- AlicenseNot gradedqualityCmaintenanceRun the full agency SEO loop from your AI assistant: Search Console insights, prioritized actions, article generation, CMS publishing, indexing, and performance measurement.MIT
- AlicenseNot gradedqualityCmaintenanceConnects AI assistants to live site data so they can answer questions about crawling, indexing, Search Console performance, keyword gaps and how AI engines describe a brand. Runs technical audits, keyword research, content workflows and client reports directly from the chat.MIT
- AlicenseNot gradedqualityCmaintenanceLets AI assistants query stored Search Console performance, crawl diagnostics, content gaps, decay, cannibalization, and internal-link opportunities, and optionally mark existing SEO to-dos done or ignored.MIT
- -licenseNot gradedqualityCmaintenanceEnables AI assistants to perform comprehensive SEO and GEO measurements, including site audits, keyword research, ranking tracking, and brand visibility analysis across search engines and generative AI platforms.-
Glama MCP Gateway
Add one secure layer between your agents and this server.