accept-money-page-suggestion
ChatGPTAccept a money page suggestion, which adds the page as a money page. Use it after list-money-page-suggestions when the user agrees the page earns them money. Use dismiss-money-page-suggestion when the page does not earn them money. A suggestion of kind gap names a page that did not exist yet, so accepting it needs the url the user built, and RankYak checks that url answers on the project's domain before accepting. Pass a description to replace the suggested one, because that sentence decides which articles link the page. Refuses a suggestion that already has an answer. Returns the new money_page_id, url, and description.
add-keywords
ChatGPTAdd keywords to a project. The import runs in the background as a keyword research run. RankYak enriches each keyword with search volume and difficulty, groups it with the keywords that share its search results, and files it under a topic. An import of at most 4 phrases usually lands within 10 seconds, so the call waits up to 15 seconds for it. When the project reviews research, the run's new clusters wait for approval instead. Use it for phrases the user named or that a tool returned. Use start-keyword-research when they cannot name the phrases, because a keyword invented to answer a question usually has no searches and stays in the project for good. Pass cluster=false when each phrase deserves its own article, such as the same service in different cities. Every keyword then becomes a cluster of its own with its primary locked. With cluster=false, detach_existing=true also moves the keywords the project already tracks out of their current clusters. That move lands at once and is not part of the run's review. RankYak deletes a cluster that the move leaves empty, together with its unstarted content plan item. An import that joins an existing cluster may become its primary keyword when it has more searches. Adding a keyword schedules nothing; plan-keyword puts its cluster on a date. Returns what the call received, what is importing, which names the project already tracked, which are new, the run's id and status, and the project's keywords for the names. Volume and difficulty are not known for a new keyword until enrichment finishes.
apply-page-audit-fix
ChatGPTApply a page's suggested fix, which list-page-audits carries with the page; only the fields the suggestion has change. For a page whose article_id is set, the suggested heading, meta title, and meta description update the RankYak article after RankYak snapshots a revision, and the change republishes through the integration. For any other page RankYak cannot change the site, so applying records that the user carried the suggestion into their own CMS, and the next crawl verifies it landed. The tool refuses a page without a suggestion. Findings the fix answers resolve at once and the page moves to the fixed list. Returns applied, article_id, article_updated, and a note saying which of the two cases happened.
approve-keyword-research
ChatGPTApprove a keyword research run that waits for review. This is the same action as the Research tab in the app. Until you approve a run, its keywords, clusters, and topics count nowhere. Approval lands everything the run proposed, except the clusters you archived with update-keyword-clusters during review. Read the run with list-keyword-research first. You can approve only a run with status in_review. A run that is still researching is not finished yet. An approved or failed run is settled. On a site that has no article yet, approval also builds the first content plan and starts today's article at once. That first plan replaces the unstarted items already on the calendar. While auto-publishing is on, that article publishes through the site's channel when it is ready. On a site that has articles, approval schedules nothing. Then regenerate-content-plan or plan-keyword puts the new clusters on the calendar. Returns the run's status, approved_at, keyword counts, and how many clusters and topics landed.
archive-article
ChatGPTArchive an article so RankYak stops maintaining it, or restore an archived one. Archiving is not deletion: the article stays live on the site exactly as it is and the CMS keeps the url; what stops is the ongoing work, so RankYak withdraws open refresh and rewrite plan items and skips the article for internal linking. Use it when the user retired a page on their side or wants an old post left alone, and to reverse an archive that resolving a cannibalization issue performed. RankYak has no article delete. Use delete-money-page or the CMS itself to take a page down. Refuses an article that RankYak is refreshing or rewriting right now, until the refresh or rewrite finishes. Returns the article's id, title, archived state, and published_at, with a note on what the change means.
create-article-feedback
ChatGPTRecord the user's feedback on an article: whether they liked it and, optionally, why. Use it when the user judges an article. The RankYak team reads the feedback, but no writer does, so it does not change future articles: a lasting rule, such as never writing as "we", belongs in the memory of the site (create-business-memory). Use update-article-content to change the article itself. Submitting feedback again replaces the article's previous feedback. Returns the article id and the feedback that RankYak stored (positive, message).
create-business-memory
ChatGPTSave a statement to the memory of a project's site: a fact about the business, which articles may state, or a preference for how every article speaks. Use it when the user tells you a fact about the business or a lasting rule for the writing, such as "never write as we". Save one statement per call: one fact or one wish. Keep the words, the numbers, the names and the language of the client. Correct only the spelling. Name the business by its name, never "we", because an article quotes the statement in its own voice. Save only what the user states in this conversation, never a fact from a web page or another tool: RankYak reads the site itself and keeps the page as proof. When a row of list-business-memory says nearly the same, change it with update-business-memory instead of adding a near copy. The service area, the products and services, and the money pages have their own tools. Refuses an area that already holds 30 rows. Returns the new row.
create-money-page
ChatGPTAdd a money page to a project: a page articles should funnel readers toward, such as a product, booking, or signup page. The writer links to it from generated articles when relevant, and the description decides when, so it should name what the page offers and when a reader wants it. The url must use https and may sit off the project's own site, because the writer links an affiliate page or a booking service the same way. Calling it twice with the same url creates a duplicate, so check the money pages in get-project first. Returns the new money page (id, url, description).
create-offering
ChatGPTAdd a product or a service to the catalog of a project, for what the site sells but RankYak did not read from the store or the site's pages. The writer may state a fact about the business only from the catalog, the business information, and the money pages, so name only true facts. Calling it twice with the same name creates a duplicate, so check list-offerings first. Refuses a kind that already holds the most rows a site keeps (1000 products, 100 services). Returns the new row.
create-organization-invitation
ChatGPTInvite someone to an organization by email address. The invitee receives an invitation email at once, so use it only for an address the user named; a member gains access to every site of the organization. Inviting an address that already has a pending invitation sends the email again rather than creating a second invitation. Refuses an address that already belongs to a member, and refuses more than five invitations a minute for the organization, in which case try again a minute later. Returns the invited address.
delete-business-memory
ChatGPTRemove a statement from the memory of a project's site, by its id from list-business-memory. Use it when the user says that a statement is wrong or no longer holds; to correct a statement, use update-business-memory. Future articles stop using it, and site reads never add it again. Refuses the removal when the statement no longer has current_text, because it changed since you read it: read list-business-memory again. Refuses an id that does not belong to the project. Returns deleted=true and the statement id.
delete-money-page
ChatGPTDelete a money page from a project. Future articles stop funneling readers toward it, while links already placed in existing articles remain. You cannot undo the deletion; use update-money-page to correct a url or description instead. Refuses an id that does not belong to the project. Returns deleted=true and the money page id.
delete-offering
ChatGPTDelete a product or a service that the user added to a project's catalog (source manual). Future articles stop using it, while existing articles keep what they say. You cannot undo the deletion. Refuses a row that RankYak read from the store or the site, because the next sync or scan can add it again; use update-offering with hidden=true to keep such a row out of articles. Refuses an id that does not belong to the project. Returns deleted=true and the offering id.
delete-organization-invitation
ChatGPTRevoke a pending invitation, so its link no longer works. RankYak does not notify the invitee, and existing members are untouched; removing a member happens in the RankYak app. Use list-organizations for the invitation id. Refuses with "Invitation not found." when the invitation is already accepted or revoked. Returns the revoked address.
disconnect-gsc
ChatGPTDisconnect Google Search Console from a site. This deletes the stored Google tokens and the configured property, so ranking and traffic data stop updating and the site audit loses its click-through signal. Reconnecting signs in with Google again and happens in the RankYak app. Use update-gsc-site-url instead when the account is right and only the property is wrong. Refuses when Search Console is not connected.
disconnect-integration
ChatGPTDisconnect a site's CMS integration, whichever type it is. Articles stop publishing until a new integration connects. Articles already on the CMS stay live. RankYak deletes the stored credentials at once. It keeps a record of the integration only while published articles still refer to it. Use it before a different platform connects, because a site holds one integration at a time. The tool refuses when the site has no integration.
dismiss-cannibalization-group
ChatGPTDismiss one consolidation plan set: the user keeps every page as it is, and each issue in the set moves to the handled list and never reopens, however often detection sees the competition again. Use it for a plan set from list-cannibalization-issues; dismiss-cannibalization-issue is for a single watch-list issue, and resolve-cannibalization-group is for a set the user does consolidate. Dismissal is permanent. Returns dismissed.
dismiss-cannibalization-issue
ChatGPTSet a single cannibalization issue aside: the evidence stays on record, and the issue moves to the handled list and never reopens, however often detection sees the competition again. Use it for a watch-list overlap the user wants to stop watching, or for pages that serve different intents despite the shared queries; for a plan set, dismiss-cannibalization-group dismisses all of its issues together, and resolve-cannibalization-issue is for an issue the user consolidates. Refuses an issue that is already resolved. Returns dismissed.
dismiss-money-page-suggestion
ChatGPTDismiss a money page suggestion, recording that the page does not earn the business money so RankYak stops proposing it. Use it after list-money-page-suggestions when the user rejects a page. Use accept-money-page-suggestion when they agree. Nothing on the site changes, and the page keeps the links it already has. Refuses a suggestion that already has an answer. Returns dismissed=true and the suggestion id.
dismiss-page-audit-fix
ChatGPTSet a flagged page aside: its findings stay on record, and the page moves from the attention list to the fixed list, with or without a generated suggestion. Use it when the user would rather leave a page as it is, or to clear a pending suggestion the user does not like, after which generate-page-audit-fixes produces a fresh one. Use apply-page-audit-fix instead when the suggestion should take effect. Refuses a page whose fix is already applied, unless list-page-audits marks it low_ctr_after_fix. Returns dismissed.
generate-article-header-image
ChatGPTGenerate a new header image for an article and set it as the header. This is the same flow as the header menu in the editor. It makes the one image that sits above the article. It never places images inside the article body. Use set-article-header-image instead to reuse an image the library already holds. Each generation spends one of the article's image generations, so the call needs confirm=true. Leave instruction empty to let RankYak write the brief from the article, which is what the editor does by default. Pass instruction only when the user described what the image should show. Describe the subject only, because RankYak applies the project's image style and colors itself. A library image that the new image replaces stays in the library. An article written before the image library may still carry a header of its own, and the new image deletes that older header for good. The tool refuses a project on the older image system, a project with images turned off, and an article with no generations left. A refusal charges nothing. Returns the new image's id and url and the article's generations_remaining.
generate-page-audit-fixes
ChatGPTQueue AI-written fix suggestions for flagged pages: a meta title, an on-page heading, and a meta description that answer the page's findings. A suggestion changes nothing on the site until apply-page-audit-fix applies it. It requires Site Guard, and RankYak meters nothing. One page id is a choice, so the tool refuses it with the reason when the page already has a suggestion waiting or no open findings, and it regenerates the suggestion of a page with an applied fix; several ids are a sweep that skips pages already generating, suggested, or applied, except an applied page that list-page-audits marks low_ctr_after_fix. Generation takes about a minute per page in the background. Returns queued (how many pages RankYak queued) and selected (how many the call named); list-page-audits has the suggestions once they are ready.
get-article
ChatGPTGet one article in full: title, meta title and description, keyword, integration, header_image_url, the searches that show the live page in Google, and its publish state (published_at, republished_at, error_at, external_url, generating_status, has_unpublished_changes). Use it after list-articles when one article is the subject. Use get-article-seo-analysis for the score. Use get-article-research for the sources. The tool includes the body only on request, because it is large: content=markdown returns the source text (the form update-article-content takes back), and content=html returns the rendered markup. With a body included, the result omits summary and achievements. A non-null error_at means the last publish failed and publish_error carries the CMS's reason. While the article is generating, generation_step and generation_progress say how far along the generation is. restore-article-revision undoes edits made through update-article-content and update-article-meta.
get-article-research
ChatGPTGet the research that RankYak wrote an article from: the search intent, article type, and target word count chosen for it, the keyword variations its cluster covers, the pages that ranked for the primary keyword at research time (urls and titles, not their content), and the related questions and searches. Use it to trace where a claim in an article came from, or to see what RankYak wrote the article to compete with; the project's business details in get-project are the other source an article draws on. Returns those fields for one article; the research of the article must be complete, so an article still in its first generation step has none yet.
get-article-revision
ChatGPTGet one earlier version of an article: what it looked like before a refresh, rewrite, edit, or restore changed it. Use it to answer what a refresh changed, and to read a revision before restore-article-revision replaces the current article with it; list-article-revisions gives the dates and reasons. The default content=diff compares the revision with the article as it stands now and is far smaller than two full bodies; content=markdown returns the whole old body. Returns the revision's id, reason, title, created_at, and the requested content.
get-article-seo-analysis
ChatGPTGet the SEO analysis the article editor shows: the score out of 100, the three categories with their points, the checks behind them, and the content statistics (word count against target, headings, paragraphs, keyword density, internal and external links). Use it for questions about an article's score or on-page SEO. Use get-article for the text itself. Read points for what each category is worth on the total, and read checks for what each one counts and the range that scores in full, because the raw category scores share no denominator. The checks follow the rules that RankYak writes articles by. The score is an on-page gauge, not a measure of quality or ranking: padding an article to reach the length target costs readability. Returns the analysis for one completed article.
get-audit-page-tree
ChatGPTBrowse one level of the site's page tree, built from the pages RankYak discovered on the user's website. Use it to see which sections and pages exist before composing a selection with update-audit-selection; get-site-audit shows the selection already saved. Each call returns the immediate children of one path, each with its page count and whether it is itself a page, has children beneath it, or both, and is_page for the requested path itself. The homepage is the root path and never appears as a child. Call again with a child's path to go one level deeper.
get-content-plan
ChatGPTGet a project's content plan: the scheduled article writes, each with the cluster it covers named by its primary keyword, the article once written, and the scheduling state. One item is one cluster, which is one article. Defaults to items from the start of the current week (Sunday) onward; pass from and to dates to change the window. A non-null blocked_at means the site's monthly article cap refused the write, which only a legacy Starter site carries; the item retries and writes once the month turns over. Returns the items with their ids, which feed schedule-keyword, reschedule-content-plan-item, and remove-content-plan-item. Also returns open_dates: the days that the site writes on and that can still take an article, through the last item or the to date, and at most a year ahead. Without a to date, every write day after the last item is also free, so a plan with no items leaves every write day free. Read open_dates before you say that the calendar is full.
get-integration-options
ChatGPTList what the project's connected CMS offers to choose from, read live from the CMS: Webflow collections (pass collection_id to also get that collection's fields for the mapping), Ghost authors, Duda sites, WordPress and WordPress.com categories, tags, and users, Shopify blogs, or Wix members. Use it before you change the integration settings, which take the ids this returns. Requires a connected integration (get-project shows it); Lovable, Replit, and Next.js have nothing to choose. Returns the integration type and the lists keyed by kind.
get-keyword-cluster
ChatGPTGet one keyword cluster in full: every keyword variation with its metrics, starred and archived state, and the pages on the site already ranking for it, plus the topic above it, the article written for it, and its content plan item. Pass cluster_id, or keyword_id to reach the cluster a keyword sits in, which answers a question about a single keyword, because the keyword is a phrase and the cluster is the article. Read this when the listing is not enough, or when a cluster gets no article: blocked_by names the one reason the planner cannot schedule it, or is null when it could. Refuses a keyword that is not in a cluster yet, because keywords join a cluster when their research run finishes.
get-project
ChatGPTGet one site in full: its settings, the article settings (videos and images), the CMS integration (type, whether it is connected, and whether its setup is finished), the Search Console connection, the developer zone switches (feed, webhook, API), the money pages, the backlink exchange state, the site level and its XP, whether RankYak read the site and when, content counts (keyword clusters, articles, money pages), and the setup guide's step states (null once the user dismisses the guide). Use it before update-project-settings or update-article-settings to see current values, and before publish-article to check the integration. The tool never returns the feed URL, webhook URL, and API key; the user reveals them in the RankYak app under Settings, Integrations, Developer zone. article_schedule_time is in the site's timezone. Money page ids feed update-money-page and delete-money-page.
get-project-statistics
ChatGPTGet a site's performance figures, as the Home page in RankYak shows them: search impressions and clicks from Google Search Console over the past 30 days against the 30 days before, the published articles with the most impressions, keyword cluster coverage (clusters that already have an article or a ranking keyword, out of the cluster pool), and backlink counts. The backlink counts cover only the links that the site received through the RankYak backlink exchange, so they are not the site's total backlinks: links from elsewhere on the web are not counted. Use it for questions about traffic, rankings, and progress. Use get-project for settings and connection state. search_performance is null until Search Console is connected; gsc_connection_status says whether it is, and connecting happens in the RankYak app.
get-recent-activity
ChatGPTGet what RankYak did for a project's site lately, how Google shows the site, which articles come next, and what holds the site back until the user acts. These are the facts of the opening line on Home. Use it when the user asks what RankYak did, such as "what did you do this week", or what happens next. The work covers the time since the user last opened the site, at most 30 days, or the last 7 days when that visit is unknown.
get-search-console-breakdown
ChatGPTRead a site's Google Search Console numbers for one period, live from Google, grouped by page, query, country, device, or search appearance. Each row holds the clicks, the impressions, the CTR, and the average position, set against an earlier period. Sort by a total, such as clicks, or by a change, such as impressions_gained or clicks_lost. Use a change sort to find what caused a rise or a drop: Google lists rows by clicks, so a page with many new impressions and few clicks shows up only in a change sort. A comparison ends on the last day that Google finished counting, so it never sets a partial day against a full one; get-search-console-trend reads the later days. The filters and the search type work as in get-search-console-trend. A page row says whether RankYak wrote the page or links to it, and a query row names the article that targets it. Returns at most 50 rows.
get-search-console-index-status
ChatGPTRead how Google indexes a site, live from Google Search Console. It returns the sitemaps of the site's property with the time that Google last read each one, and the URL Inspection result of up to 5 pages: whether Google indexed the page and why not, a robots.txt or noindex block, the last crawl, and the canonical that Google chose. Use it when the user asks whether a page or an article is indexed, and when impressions fall toward zero. Without urls, it inspects the homepage. Each url must sit inside the site's Search Console property. The result is the state that Google recorded at its last crawl, so it can lag the live page by days.
get-search-console-trend
ChatGPTRead a site's Google Search Console numbers over time, live from Google. Each point holds the clicks, the impressions, and the average position of an hour, a day, a week, or a month, set against the same point of an earlier period. Use it when the user asks how search traffic moved: a rise, a drop, one day, or today so far. Google counts days in Pacific time and finishes a day about two days later, so each point says whether Google finished it. An unfinished point runs low and stays out of the totals. Grain hour sets today so far against the same hours of an earlier day. Filter by page, query, country, device, or search appearance, and choose a search type such as image or discover. The result also lists what RankYak did on each day, such as an article that it published or optimized. To find which pages or queries caused a change, use get-search-console-breakdown. When the complete days fall toward zero, read get-search-console-index-status.
get-site-audit
ChatGPTGet a project's Site Guard overview: whether the site carries Site Guard, the saved page selection and how many pages it covers, the meta title boilerplate the audit ignores, whether Search Console is connected, the latest crawl with its progress, and the open finding counts per check. Use it first for any Site Guard question, and for the counts on a site without Site Guard, which gets the counts and nothing more; list-page-audits has the flagged pages themselves. The response carries has_site_guard, enabled (a non-empty selection), selection, meta_title_boilerplate, has_gsc_connection, latest_run, totals (pages_audited, pages_flagged, open_findings per check, fixes_ready, open_cannibalization_issues), bounds (the recommended meta title and description lengths), selected_pages_count, money_page_coverage (how many of the site's own pages link each money page; a page marked external lives off the site on purpose, and on_site false means the crawl expected the page and did not find it, which is usually a typo in the URL), and cannibalization_audit.running.
list-article-images
ChatGPTList the images in a project's library, newest first: the headers, in-article images, and infographics RankYak generated, the files the user uploaded, and the catalog photos RankYak copied from the user's site or store. Use it to find an image id for set-article-header-image, or to see which article an image is for. Search matches the name, the alt text, and the instruction that RankYak generated the image from; type and source narrow the list. Returns one page of images with id, name, alt, type, source, url, article_id, and created_at, plus pagination. A project on the older image system has no library and returns an error.
list-article-revisions
ChatGPTList an article's content revisions, newest first: the snapshots taken before a refresh, rewrite, manual edit, or restore. Use it to find the revision to inspect with get-article-revision or to restore with restore-article-revision; an article without refreshes or edits has none. Returns a paginated data list of id, reason, title, and created_at; the list does not include the content of a revision.
list-articles
ChatGPTList a project's articles, newest first, with their generation and publish state. Use it to find an article's id and to answer which articles are published, still generating, failed to publish, or archived; get-article returns one article in full. Filter by published, unpublished, needs_republish (published articles that changed after their last publish, by an edit, a refresh, or rebuilt internal links), or archived. Search by title or keyword. Returns a paginated data list where each article carries id, title, generating_status, published_at, republished_at, has_unpublished_changes, error_at, archived_at, created_at, the keyword it targets (id, name, search_volume, keyword_difficulty), and its content plan item (id, publish_at). The ids feed get-article, update-article-content, publish-article, and list-article-revisions.
list-business-memory
ChatGPTList what RankYak knows about the business behind a project's site, by area. Facts (business, audience, services, policies, proof, contact) say what is true about the business: the writer states facts about the client only from these rows, the offerings, the money pages, and the business information. Preferences (voice, avoid) say how articles speak, such as "never write as we". A row with source website comes from a page of the site, with the page and a verbatim quote in sources; a row with source user is what the user told RankYak. Read it to answer why an article says or avoids something. Returns the state of the last site read (read_at, error) and the rows with id, area, text, source, sources, author, confirmed_at and updated_at.
list-cannibalization-issues
ChatGPTList a project's cannibalization findings in the units the user acts on. It requires Site Guard, because the issue detail is what Site Guard adds to the count get-site-audit reports. The open filter (the default) returns the consolidation plan in data: each entry is one plan set, the pages the audit judged to serve one search demand, merged from every duplicate-classified detection that shares a page, with its pages, a proposed winner_page_id, a winner_reason, and the underlying issue_ids. Alongside it come watch, the overlaps classified as bleed, which need no action, and unjudged_detections, the number the next run still has to judge. The handled filter returns in data the groups already resolved or dismissed, each with its status. A page appears in exactly one plan set, so a set's issue_ids feed resolve-cannibalization-group and dismiss-cannibalization-group, and a watch entry's id feeds resolve-cannibalization-issue or dismiss-cannibalization-issue. Each page carries its url, is_winner, metrics, and the article behind it when RankYak wrote one.
list-keyword-clusters
ChatGPTList a project's keyword clusters, the unit RankYak plans and writes in: one cluster is one article, headed by its primary keyword and holding the variations that share its search results. Read this when the question is what the site covers, what it could cover next, or what a given article is for. Read list-keywords for the variations inside the clusters. Filter with status: schedulable clusters are the ones the planner could pick up, planned and written ones already have a date or an article, awaiting_review ones belong to a research run that still waits for review and count nowhere yet, and all is everything except archived. Returns a page of clusters with their primary keyword and metrics, a preview of variations, topic, article, content plan item, and the archived and awaiting_review flags. Cluster ids feed get-keyword-cluster and update-keyword-clusters.
list-keyword-research
ChatGPTList a project's keyword research runs, newest first, with what each found. Read this when a project looks empty or a keyword the user expects is missing: until a run lands, its clusters count nowhere and list-keywords does not show their keywords. A run with status researching is not finished yet; one with status in_review waits for the user, and for it the response carries waiting_for_review with the proposed clusters and their ids, the topics, and the strongest keywords, so the user can see what approving would land. Returns a page of runs with id, type, status, keyword counts, and timestamps, and reviews_research, which says whether the project reviews every run; a goal research (type context) waits for review regardless. Run ids feed approve-keyword-research; cluster ids feed update-keyword-clusters.
list-keywords
ChatGPTList a project's approved keywords with search volume, difficulty, intent, starred state, the cluster each sits in, that cluster's topic, content plan item, and article, and the pages on the site already ranking for it. Each row is one variation, not one article: rows that share a cluster share one article, and the cluster block names it. Read list-keyword-clusters when the question is how many articles the site is worth. Keywords appear here only once their research run lands, so an empty list usually means a run is waiting in list-keyword-research. Filter with type: starred, opportunity (high potential), planned (the cluster has a content plan item), or unplanned. Returns a page of keywords; their ids feed plan-keyword, schedule-keyword, update-keywords, and get-keyword-cluster.
list-money-page-suggestions
ChatGPTList the pages RankYak read on the project's own site and believes earn the business money, waiting for the user's answer. Use it when the project has no money pages yet or the user asks which pages RankYak found. Answer each suggestion with accept-money-page-suggestion or dismiss-money-page-suggestion. A project whose site RankYak did not crawl yet has none, and so does one that already named its money pages. Returns a data list of money_page_suggestion_id, kind (an existing page, or a gap RankYak thinks is missing), url and path, suggested_description, reason, and confidence between 0 and 1, plus money_pages_named.
list-offerings
ChatGPTList what a project's site sells: its products, collections, and services, 50 per page, sorted by name. The writer may state a fact about the business only from these rows, the business information, and the money pages, so read them to answer what the site sells or why an article names a product. RankYak reads a Shopify or WooCommerce store from the store and every other site from its own pages; the user adds the rest. Filter by kind and search by name. Returns the catalog state (source, synced_at, error) and a paginated data list where each row carries id, kind, source (shopify, woocommerce, website, or manual), name, url, price, in_stock, description, facts, hidden (kept out of articles), and edited. The ids feed update-offering and delete-offering.
list-organizations
ChatGPTList the organizations the user belongs to, each with its sites, its members (name, email, and whether they own the organization), and its pending invitations. Use it to find an organization_id for update-organization and create-organization-invitation, and an invitation id for delete-organization-invitation. is_owner says whether the user owns the organization. Every member may manage every site of the organization.
list-page-audits
ChatGPTList a project's audited pages with their open findings, fix status, suggested fix, and Search Console metrics, highest impressions first. Use the default needs_attention filter for the pages with open findings not yet acted on, plus the applied pages whose click-through rate is low on Search Console data from after the fix. Use fixed for the other pages whose fix is applied or dismissed. It requires Site Guard on the site, because the page detail is what Site Guard adds to the counts get-site-audit reports. Each row carries the page id, url, meta_title (the title tag as crawled), article_id and article_title when a RankYak article stands behind the page, fix_status, fix_applied_ctr (the click-through rate when the user applied the fix), low_ctr_after_fix (true for such an applied page), suggested_title (the heading), suggested_meta_title, suggested_meta_description, and open_findings with their type and data. Page ids feed generate-page-audit-fixes, apply-page-audit-fix, and dismiss-page-audit-fix.
list-projects
ChatGPTList the sites the user may manage, across every organization they belong to. Start here: every other tool except the organization tools takes a project_id from this list. Returns, per site, its id, name, domain, organization_id, whether it is active (a deactivated site is read-only), and whether onboarding is complete. Use get-project for a site's settings and connections.
list-topics
ChatGPTList a project's topics, the top level of the topical map, with the cluster and keyword count of each. Read this for the shape of a site. Read list-keyword-clusters for the work itself, because each cluster under a topic is one article. The list leaves out archived topics unless include_archived is true. Returns each topic's id, name, is_core_topic, keyword_research_status, archived_at, clusters_count, keywords_count, and awaiting_review, which is true for a topic from a research run that still waits for review, so it counts nowhere until the run lands. Topic ids feed update-topic and the topic_id filter of list-keyword-clusters.
merge-keyword-clusters
ChatGPTMerge clusters: move every keyword of the given clusters into the target and delete the emptied clusters. The target keeps its primary keyword, and the moved keywords become its variations, pinned so research never moves them back. Use it when two clusters are one article's worth of demand. Use move-keywords to move single variations. The merge removes an unstarted content plan item of a merged cluster. Refuses a cluster that has an article or whose article RankYak is writing, an archived cluster, a cluster whose research run is not approved, and a target that is archived or listed among the clusters to merge. Cluster ids come from list-keyword-clusters. Returns the target cluster with every keyword it now holds.
move-keywords
ChatGPTMove keywords into another cluster, or give each a cluster of its own. Pass target_cluster_id to move them into that cluster, which keeps its primary keyword. Pass own_cluster=true to split each keyword into a single-keyword cluster under the same topic, which is how a variation that deserves its own article gets one. The call pins a moved keyword to the cluster you put it in, so keyword research never moves it again; a keyword split out also becomes the locked primary of its new cluster. The call deletes a cluster left empty along with its unstarted content plan item, and a cluster whose primary left elects a new primary. Refuses the primary keyword of a cluster that has an article or a content plan item (promote another keyword first, or use merge-keyword-clusters), a keyword or target whose research run is not approved, an archived target, and splitting out an archived keyword. Keyword ids come from list-keywords, and cluster ids come from list-keyword-clusters. Returns the keywords that moved with their new cluster ids, and the ones that already sat where you asked.
plan-keyword
ChatGPTPut a keyword's cluster on the content plan on a date of its own. The call creates the calendar slot. The article covers the whole cluster, so any keyword in the cluster plans the same article. Use it for "plan an article about X" and "add another article next week". Use schedule-keyword when an existing slot should hold this keyword instead, because that swaps the slot's keyword and adds no day. Omit publish_date to take the first free day. A date is available while the site's article time on that day is still ahead, because the article must be finished by that time. The tool refuses a keyword that is archived or not in a cluster yet. It refuses a cluster or a topic that is archived. It refuses a cluster whose research run is not approved, and a cluster that is already planned or written. The refusal names the reason. The tool refuses a day that already holds an article unless replace_occupant is true. Then it removes the article planned on that day, and that removal needs confirm=true. Returns the new content plan item with its id, the primary keyword that RankYak writes the article for, and publish_at.
promote-keyword
ChatGPTMake a keyword the primary keyword of its cluster, which is the keyword that RankYak writes the cluster's article for. The call locks that choice, so keyword research never elects a different primary by search volume. Research may still add variations to the cluster. Use it when the highest-volume keyword is not the one the article should target, such as when a city-specific phrase should lead. When the cluster already has an article or a date on the content plan, the article's focus keyword and every future rewrite follow the new primary. Its title and URL stay as they are. That change needs confirm=true, passed only after the user agreed to it. A cluster without an article or a planned date needs no confirmation. The tool refuses an archived keyword and a cluster whose research run is not approved. Keyword ids come from list-keywords. Returns the cluster with its new primary. Call update-keyword-clusters with primary_locked=false to hand the choice back to RankYak.
publish-article
ChatGPTPublish a completed article that is not yet published to the project's CMS integration. Use publish-article-changes for an article that is already published and has edits to push. Use publish-article-to-webhook to deliver to the project's own webhook instead of a CMS. RankYak queues the publish, which runs in the background; published_at and external_url in get-article show when it finished. Refuses with the reason when the article is not completed, is already published, or the project has no integration whose setup is finished. Returns the article id and queued=true.
publish-article-changes
ChatGPTPush the unpublished edits of an already-published article to the project's CMS integration and webhook. Use it after update-article-content, update-article-meta, or restore-article-revision when has_unpublished_changes is true. Use publish-article for an unpublished article, and use publish-article-to-webhook for a webhook-only delivery. RankYak queues the republish, which runs in the background. Refuses when the article is not completed or has no edits newer than its last publish. When the project has neither an integration nor a webhook, RankYak marks the edits as published without sending them anywhere. Returns the article id, republish_queued, and webhook_queued.
publish-article-to-webhook
ChatGPTSend a completed article to the project's webhook, without touching the CMS integration. Use this when the user's own system receives articles through the webhook (Zapier, a custom endpoint, an app builder without an integration) and they want this article delivered now, for example to deliver one again or to deliver an article that RankYak wrote before the user enabled the webhook. Requires an enabled webhook with a URL (update-developer-settings). Delivery runs in the background. Use publish-article to publish through the CMS integration. Use publish-article-changes to push edits of an already-published article to every channel the project has.
read-help-article
ChatGPTRead one RankYak help center article in full, as Markdown. Use it to answer questions about how RankYak works from the official documentation instead of memory, with a slug from the help center index or from search-help-articles. Use search-help-articles when you do not know a slug. Returns the article's slug, title, description, public url, and markdown; the url is public, so you can link it in an answer.
regenerate-content-plan
ChatGPTRun RankYak's planner: it selects the best unwritten clusters and fills the upcoming schedule with them, one per writing day. Use it whenever the user wants the plan filled or rebuilt with the best keywords. Use plan-keyword or schedule-keyword to place one article by hand. The tool replaces every unscheduled content plan item (unstarted upcoming items, plus pending refreshes and rewrites), and RankYak cannot restore the removed items, which is why the tool takes confirm. An error leaves the plan unchanged. Returns regenerated and the number of upcoming items on the plan.
remove-content-plan-item
ChatGPTTake a planned article off the calendar. The keyword's cluster returns to the pool, so the planner may pick it again later, and nothing else on the calendar moves. Use this when the user wants a planned article gone. Use reschedule-content-plan-item when they only want a different date, and use schedule-keyword when they want another keyword to take over the slot. This tool cannot remove an item whose article already started generating, or is already published. Item IDs come from get-content-plan. Returns the removed item's id, date, and keyword.
reschedule-content-plan-item
ChatGPTMove a planned article to another date, the same as dragging it across the calendar; the item keeps its keyword and only the date changes. Use it for "move this article to next week". Use schedule-keyword to change the keyword a day holds, and use remove-content-plan-item to take an article off the calendar. A date is available while the site's article time on it is still ahead. Refuses an item whose article already started generating or is published. The tool refuses a day that already holds an article unless swap_with_occupant is true, which trades the two days. Item ids come from get-content-plan. Returns the item's id, its primary keyword, moved_from, and the new publish_at; only this item moves, so the rest of the calendar stays where it is.
resolve-cannibalization-group
ChatGPTResolve one consolidation plan set: mark its issues solved and, with archive=true, archive every RankYak article in the set except the winner. Use it for a plan set from list-cannibalization-issues; resolve-cannibalization-issue is for a single issue that stands alone, because a page may sit in several detections and resolving them one by one can archive a page another detection keeps. Archiving needs the user's agreement because the pages stay live but lose their maintenance, so the tool refuses without it. With archive=false RankYak marks the set solved and nothing changes, the path for a consolidation that the user handled outside RankYak. Returns resolved, archived_articles (the losing urls), and redirects_to_create, the 301 redirects the user sets up in their own CMS, because RankYak cannot create redirects itself.
resolve-cannibalization-issue
ChatGPTResolve a single cannibalization issue: mark it solved and, with archive=true, archive every losing RankYak article behind the chosen winner. Use it for an issue that stands alone, for example a watch-list overlap the user decided to consolidate anyway; for a plan set, resolve-cannibalization-group resolves all of its issues together, because a page may sit in several detections and resolving them one by one can archive a page another detection keeps. Archiving needs the user's agreement because the pages stay live but lose their maintenance, so the tool refuses without it. Archiving also needs winner_page_id. With archive=false RankYak marks the issue solved and nothing changes. The tool refuses an issue that is already handled. Returns resolved, archived, and redirects_to_create, the 301 redirects the user sets up in their own CMS.
restore-article-revision
ChatGPTRestore an article to an earlier revision. RankYak first snapshots the article's current state as a new revision, so restoring that snapshot can undo the restore itself; get-article-revision shows a revision before you restore it. Use it to undo a refresh, rewrite, or an edit made through update-article-content. Refuses while the article is still generating. The restored text does not reach the live site by itself; publish-article-changes pushes it. Returns the article id, the restored revision id, and the id of the snapshot taken.
schedule-keyword
ChatGPTChange which keyword cluster an existing content plan item holds; the item keeps its date and only its keyword changes, and RankYak writes the article for the cluster's primary keyword. Use it when the user wants a keyword to take over a slot that already exists. Use plan-keyword to add a day, and use reschedule-content-plan-item to move one. Refuses an item that already has an article or whose article time is in the past, and a keyword that is archived, not in a cluster, in an archived cluster or topic, in an unapproved research run, or already planned or written. Item ids come from get-content-plan, and keyword ids come from list-keywords or list-keyword-clusters. Returns the item's id, the primary keyword it now holds, and its publish_at.
search-help-articles
ChatGPTSearch the RankYak help center, the official documentation, by meaning. Use it for questions about how RankYak works (features, settings, integrations, publishing) when you do not know an article slug; read-help-article returns a whole article by slug. Describe what you are looking for in a short phrase; the wording need not match the docs. Returns a data list of matching articles (slug, title, url) with the sections that matched (heading, url, content), best match first, plus degraded, which is true when the search fell back to keyword matching.
set-article-external-url
ChatGPTSet the live url of an article the user published themselves, outside any integration, and mark it as published. Use it after the user copied the article into their own CMS. Use publish-article when the project has an integration that can publish it. The url must be on the project's own domain and publicly reachable, because RankYak probes it with an HTTP request before saving, so the tool refuses a url on another domain or one that does not answer. Returns the article id, external_url, and published_at.
set-article-header-image
ChatGPTPoint an article at an image from the project's library as its header, or clear the header with a null image_id. Use it to reuse an existing image. Use generate-article-header-image to make a new one. Any image in the library qualifies, including one made for another article. Find ids with list-article-images. A library image that this call replaces stays in the library. An article written before the image library may still carry a header of its own. A choice made here deletes that older header for good, and clearing the header deletes it too. A new header is an unpublished change, and publish-article-changes sends it to the live site. A cleared header is not: most CMSes keep the old image until the user removes it there. Returns the article id and the header image's id and url, or null when cleared.
start-keyword-research
ChatGPTStart a keyword research run from a description of the business, the same action as Research keywords in the app: RankYak reads the description, works out the topics the site should own, and builds keyword clusters under them. Use it when the user wants coverage of something new and cannot name the keywords. Use add-keywords when they already know the phrases. Write context in the user's own words about the business and the subject, because that text is the whole input. A run takes several minutes, so the response reports that it started; list-keyword-research shows its progress. A project runs one research at a time, so nothing starts while a run is still researching or waiting for review; the response then names that run instead. A project starts at most 10 goal researches a day, counted in the project's time zone, so a call past that limit returns a validation error until the next day. A goal research always waits for the user's review once it finishes, whatever the project's auto-approve setting. Returns started and the run's id and status.
update-article-content
ChatGPTReplace an article's title, and optionally its whole body, the same way a save in the editor does. Use it when the user wants the text of an article changed. Use update-article-meta for the meta title and description. Use restore-article-revision to bring an earlier version back. Read the current body first with get-article and content=markdown. Then send the full edited body as markdown. The call replaces the body in whole. A body left out keeps the current one, and only the title changes. The article must finish generating first. RankYak snapshots the current text as a revision before the first edit after a publish, so restore-article-revision can bring the published text back. Consecutive edits share that one snapshot. The change stays in RankYak until publish-article-changes pushes it to the site. Returns the article id, its title, the id of the revision snapshot, and has_unpublished_changes.
update-article-meta
ChatGPTSet an article's meta title and meta description, the two fields search engines show in results. Use it for title-tag and snippet work. Use update-article-content for the article title and body. The call replaces both fields with what you send. An empty or omitted field clears it, and RankYak then falls back to the article title. Each field is at most 255 characters. The article must finish generating first. RankYak snapshots the current text as a revision before the first edit after a publish, so restore-article-revision can bring the published text back. Consecutive edits share that one snapshot. The change stays in RankYak until publish-article-changes pushes it to the site. Returns the article id, both fields as stored, and has_unpublished_changes.
update-article-settings
ChatGPTUpdate what a site's generated articles contain: YouTube videos, AI-generated images (a header image, images inside the article, infographics), the image style, and the image colors. Only the fields you pass change. Enabling images needs an image_style, and a site on the older image system also needs image_background_color and image_accent_color, set earlier or in the same call. While images are off, RankYak neither applies nor checks the image sub-settings you pass, so the site keeps them for when images come back on. The user configures the header image template (fonts, overlays, background photos) in the RankYak app, because it involves file uploads. Use update-project-settings for the schedule and the automation switches. Returns the settings after the change.
update-audit-meta-title-boilerplate
ChatGPTReplace the meta title boilerplate the site audit ignores when it measures meta title lengths: the branding a site adds to every title, such as "| RankYak" in "Blog | RankYak". Each entry carries its separator on the side that touches the title, so a suffix reads "| RankYak" and a prefix reads "RankYak |". The call replaces both lists in full; get-site-audit shows the current ones. The first audit detects the boilerplate itself, but a list saved here is the user's own and detection never overwrites it. A change re-checks the existing findings against the pages already crawled, so it takes effect without a new crawl. Returns the saved meta_title_boilerplate.
update-audit-selection
ChatGPTReplace the page selection the site audit crawls. The call replaces both lists in full, so pass the complete selection and not a diff. get-site-audit shows the current selection, and get-audit-page-tree shows the paths to choose from. Each path covers itself and everything beneath it. The most specific matching path decides, so "/" included with "/docs" excluded audits everything but the docs, and "/docs/api/intro" included again pulls that one page back in. The homepage flag exists because "/" covers the whole site and cannot single the homepage out. When you omit the flag, the homepage follows the path rules. A save deletes the audited pages the new selection no longer covers, together with their open findings. A page with a fix suggestion, and a page already handled, keep their history. A save starts no crawl; the weekly crawl reads the new selection. Returns enabled (whether the selection covers any page) and the saved selection.
update-backlink-exchange
ChatGPTJoin the backlink exchange, or leave it. Sites in the exchange link to each other from their published articles, so a site that never joined receives no backlinks, and the articles that RankYak writes for a site that joined may carry links to other sites; a bare backlink count cannot show that, which makes this the usual answer to "why do I have no backlinks". The tool refuses to join while the site is classified high risk, or while the site has no article published in the past 30 days with a live URL that RankYak knows; get-project reports both as backlink_exchange.eligible. The articles need no outbound links. The help article backlink-exchange explains when the first links arrive. The tool always lets the site leave, and leaving keeps the links that other sites already placed. Returns the exchange state.
update-business-memory
ChatGPTChange the text or the area of a statement in the memory of a project's site, by its id from list-business-memory. Use it when the user corrects a statement or says it in other words. Keep the words, the numbers, the names and the language of the client. Correct only the spelling. Name the business by its name, never "we", because an article quotes the statement in its own voice. A changed statement from the site (source website) becomes a statement of the user, and site reads never add the old words again. Refuses the change when the statement no longer has current_text, because it changed since you read it: read list-business-memory again. Refuses an id that does not belong to the project. Returns the saved row, which has a new id when the statement came from the site.
update-developer-settings
ChatGPTSwitch a project's developer zone channels on or off: the RSS feed of its articles, the REST API, and the webhook that receives each published article. Pass only the switches you want to change; the others keep their state. Enabling the webhook needs webhook_url; disabling it keeps the stored URL, so you can turn it back on without retyping. Use get-project to read the current state. This tool never returns the feed URL, the API key, and the webhook secret, and the user can regenerate the API key only in the RankYak app under Settings > Integrations > Developer zone. Returns the three switches and the webhook URL after the change.
update-gsc-site-url
ChatGPTSet or change the Google Search Console property a site reads its ranking and traffic data from. Use it when the connected Google account has access to several properties and the site reads the wrong one, or none yet. The Google account must already be connected. That connection happens in the RankYak app, because it signs in with OAuth. The tool refuses when no account is connected. RankYak checks the property against the connected account before it saves it, so it refuses a property the account cannot read. A changed property deletes the Search Console rows RankYak synced from the previous property. The site's performance data stays empty until the fresh sync of the new property finishes. Returns the saved property.
update-keyword-clusters
ChatGPTArchive keyword clusters, restore archived ones, or lock and unlock their primary keyword, in one call. Archiving takes a cluster off the topical map, so the planner stops considering it and RankYak writes no article for it; because one cluster is one article, this is how to drop a subject, and it beats archiving its keywords one at a time. What archiving means depends on the run the cluster came from. While its research run waits for review, archiving rejects the cluster and the run lands without it. After the run lands, archiving removes a subject that the site already tracks and also removes any unstarted scheduled article for it; restoring the cluster does not bring that content plan item back. primary_locked=false hands the choice of primary keyword back to research, which then elects the highest-volume member again; primary_locked=true keeps the current primary as it is, the same as promote-keyword on it. Refuses a cluster id the project does not own. Get cluster ids from list-keyword-clusters, or from list-keyword-research while a run is in review. Returns each updated cluster with its primary keyword, topic, archived state, lock state, and whether it still awaits review.
update-keywords
ChatGPTStar, unstar, archive, restore, pin, or unpin keywords that are already in the project. Starring marks a keyword the user approves of. Archiving depends on where the keyword sits in its cluster: archiving a variation drops that phrase and leaves the plan alone, while archiving a cluster's primary keyword archives the whole cluster, so every variation goes with it and RankYak removes the cluster's unstarted content plan item; RankYak keeps items whose article already started, and restoring does not bring a removed item back. To drop a subject on purpose, update-keyword-clusters says so directly. A pinned keyword stays in the cluster the user put it in; pinned=false lets keyword research move it between clusters again, and move-keywords is how to place it somewhere else on purpose. Refuses a keyword id the project does not own. Keyword ids come from list-keywords. Returns each updated keyword with its starred and pinned state, its role (primary or variation), its cluster id, and the archived state of the row the write touched.
update-money-page
ChatGPTUpdate a money page's url and description. Both fields replace the current values, so read the money page in get-project first and resend the field you do not want to change. A changed url must use https; you may keep an existing http url as it is. Refuses an id that does not belong to the project. Returns the updated money page (id, url, description).
update-offering
ChatGPTChange a row of a project's catalog. Only the fields you pass change. A change of the name, url, price, description, or facts marks the row edited, so the next sync or scan keeps the user's values. hidden=true keeps the row out of future articles and hidden=false brings it back; the switch alone does not mark the row edited. A Shopify or WooCommerce row takes its url, price, and stock from the store, so this refuses a changed url or price for it. Refuses an id that does not belong to the project. Returns the updated row.
update-organization
ChatGPTRename an organization. Any member may rename it, so the organization only has to be one the user belongs to. Use list-organizations for the organization_id. Returns the organization's id and new name.
update-project-settings
ChatGPTUpdate a site's settings. Only the fields you pass change; every other setting keeps its current value. Covers the general settings (name, description, audience, service area), the article schedule (timezone, time of day, weekdays), the automation switches (auto scheduling, automatic keyword research and whether its runs land without review, auto-publishing, auto-republishing), and the email notifications. The weekdays decide what the site does: a week with no write day writes nothing, and a week with no optimize day optimizes nothing. The domain, country, and language cannot change here. The user changes the language under Settings, then General, in the RankYak app. The domain and the country anchor the research already done, so to change one, the user contacts support@rankyak.com. Use get-project first to read the current values. Use update-article-settings for what generated articles contain. Returns the site as get-project does, with article_schedule_time in the site's timezone.
update-topic
ChatGPTArchive a topic, or restore an archived one. Archiving takes every cluster under the topic off the topical map at once, so use it when the user wants a whole subject gone. Use update-keyword-clusters to drop one cluster, and use update-keywords to drop one keyword variation. Because a planned article belongs to a cluster, archiving also removes the unstarted content plan items of the topic's clusters; RankYak keeps items whose article already started, and restoring the topic does not bring removed items back. Refuses a topic id the project does not own. Returns the topic's id, name, and archived state.
update-wordpress-com-integration-settings
ChatGPTChange the project's WordPress.com integration: the default categories, tags, and author new posts carry, and the draft switch. The connection itself is an OAuth grant made in the RankYak app, so there is no credential to rotate here and setup is already finished; all of these are optional defaults, and get-integration-options lists the site's categories, tags, and users with the ids this takes. Only the arguments you pass change. Requires a connected WordPress.com integration; self-hosted WordPress has its own tool. Returns the integration's type, is_connected, setup_finished, publish_as_draft, and the defaults in force.