LandingRabbit

Lists the keywords the workspace could realistically rank for in search, grouped into topics, each with suggested pages for it. These are the saved opportunities behind the workspace's Search opportunities dashboard. The only source today is SEO, based on analysis of the workspace's confirmed competitors. Each topic carries its keywords (searches per month, difficulty, intent, competitor positions and your recorded rank), an opportunity score, and page ideas. Pass topicId to get one topic with the full page angle and rationale of every idea. Saved topics can include ideas already used to create pages: check each idea's status (pending or created) and createdPageId before recommending new content. A created idea carries the id of the page it produced; a pending idea has a null createdPageId. Do not assume every returned topic still lacks a page. WHEN TO USE: any question about search traffic, keyword opportunity, competitor search visibility, or what to publish next — for example, what to write next, which keywords are winnable, where competitors rank ahead, or picking a topic by search volume or difficulty. PRESENTING AND USING RESULTS: - Present one row per topic: top keyword, totalVolume, topKeywordDifficulty, best recorded competitor rank, and number of page ideas. Distinguish pending ideas from created ideas. - yourRank is the workspace's own recorded position for a keyword; summary.competitors gives each competitor's share of the workspace's keywords. Use them to answer where a competitor ranks ahead. - Call content_gaps with a topicId before building an idea so its full angle and rationale are available. - When the user asks to create a page, pass the selected idea's suggestedTitle, pageAngle, whyRelevant, pageFormat and relevant topic keywords as materials to create_page. - A topicId request returns detail instead of a list. sortBy, limit and offset do not affect that result, but supplied arguments must still be valid. READING THE NUMBERS: - totalVolume is an estimate across the topic's keywords with near-identical keywords counted once; it is not the sum of the keyword rows. The workspace summary is computed across all topics, not just the returned page. - Topic volume covers the keyword cluster. For an individual idea's target-keyword metrics, match idea.topKeyword to the topic's keyword rows. - keywordDifficulty and topKeywordDifficulty estimate how hard it is to rank high, on a 0 to 100 scale. null means the value is not available; never read it as easy. - New keyword discovery uses a minimum monthly search volume of 100 and maximum difficulty of 60. Saved keywords remain visible when refreshed metrics fall outside those thresholds. A competitor position is recorded only inside Google's top 20, and a competitor counts as holding a keyword inside the top 10. yourRank is recorded only at position 21 or worse. Missing rank data does not prove that a competitor does not rank, and a null yourRank does not prove that the workspace does not rank: it can also mean the workspace ranks inside the top 20. - recentMonthlySearches contains up to 12 months on every response, list or detail. monthlySearchesFrom identifies the month of the first value, and the series then runs newest first. When monthlySearchesFrom is null the calendar dates are unavailable and the order is not guaranteed. - Data refreshes weekly. Members can run up to three extra refreshes a week from Search opportunities in the app. - Saved topics remain available while a refresh is running or after a refresh fails. Check refreshState and lastRefreshedAt for freshness. A null lastRefreshedAt means no completion timestamp is recorded, not that the data was never refreshed. Topics require confirmed competitors under Brand in LandingRabbit. data.status is ok, unavailable, no_competitors, running, failed or empty for lists. A topicId request returns only ok, not_found or unavailable. An error response carries no data at all. Non-ok statuses are terminal for the current request: do not immediately retry. For not_found, list current topics without topicId if needed.

content_gaps

Creates a complete page of any type, for example a landing page, blog post, help doc, or article, from the user's ideas and materials in one call. WHEN TO USE: When the user describes the page they want, for example a topic, a brief, notes, a transcript, or keywords, and wants it generated immediately. This tool: - Auto-detects the right page type from the materials - Creates a detailed value proposition plan behind the scenes - Generates the page and returns an edit URL and preview URL INPUT IS TEXT ONLY. LandingRabbit receives no files. If the user shares a file such as a PDF, slide deck, or document, extract its text yourself and pass the text as materials. DO NOT USE: When the user pastes finished copy to convert as is — use create_page_from_content. To change or recreate an existing page — use list_workspace_pages, edit_page_content, edit_page_styles, edit_page_images, and custom_design instead. If the user shared a reference image but did not say whether this is a new page or an existing page edit, ask one short clarification before creating. If the work takes longer than about 20 seconds, the reply carries an operationId instead of the result; fetch it with get_operation_result. Do not repeat a call that returned an operationId.

create_page

Creates a LandingRabbit page from finished copy pasted into the conversation. WHEN TO USE: When ready-made copy already exists — written by a human or an AI — and the user wants it converted into a structured LandingRabbit page as is. Accepts plain text, HTML, or structured copy from sources such as Google Docs or Notion. The best input is LandingRabbit-ready markdown: call get_page_format_guide for the full specification. Users who draft copy in Claude or ChatGPT before connecting can install the public skill at https://github.com/LandingRabbit/landingrabbit-skills. INPUT IS TEXT ONLY. LandingRabbit receives no files. If the copy lives in a file such as a PDF, slide deck, or document, extract its text yourself and pass the text. THE INPUT MUST BE THE PAGE COPY ITSELF. A description of a page to make, such as a brief, a topic, or instructions like "create a blog post about my call with…", is not content — pass it to create_page instead. Otherwise the request text itself becomes the page. DO NOT USE: - When the user wants new content generated from ideas and materials such as page keywords, slide decks, or meeting notes — use create_page instead - For bulk imports of multiple posts — use migrate_blog_content instead - To change or recreate an existing page — use edit_page_content / edit_page_styles / custom_design instead For reference-image work, use this only when creating a new page from ready-made copy you extracted from the image or user-provided content. If the target page is unclear, confirm new-page-vs-existing-page first. If the work takes longer than about 20 seconds, the reply carries an operationId instead of the result; fetch it with get_operation_result. Do not repeat a call that returned an operationId.

create_page_from_content

Exports every page in the workspace to a CSV (meta title/description, content as Markdown, publish state, collection, categories, tags, author). WHEN TO USE: The user wants to analyze, audit, or compare all page content (e.g. SEO cannibalization checks) or download a backup. ALWAYS use this tool to produce a pages CSV. NEVER assemble, reconstruct, or hand-write the CSV yourself from page data already in context — only this tool produces the canonical columns, ordering (most recently created first), and filename. Do not infer, rename, drop, or invent columns, and do not paste CSV contents into your reply. Show the full download URL this tool returns as plain text (do not shorten it or hide it behind a link label) so the user can copy it or use it from a terminal. The export is prepared in the background: the first call returns status "pending" with an exportId — call this tool again passing that exportId to get the download link. If the work takes longer than about 20 seconds, the reply carries an operationId instead of the result; fetch it with get_operation_result. Do not repeat a call that returned an operationId.

export_pages_csv

Fetch the result of a tool call that returned an operationId. Each call waits up to 20 seconds for the work to finish before answering. Status is one of "queued" or "processing" (call again), "completed" (the tool's result is in result), "failed" (report the message and stop), or "not_found" (the operationId is unknown here, stop and do not retry). Stop polling and tell the user the request is taking unusually long once elapsedSeconds passes 600. Safe to call repeatedly.

get_operation_result

Reads a page's current sections: section ids, types, layouts, content fields and item ids. Call it before editing a page with edit_page_content so you can reuse existing ids. Read-only.

get_page_content

Returns the LandingRabbit page format: the markdown specification for copy that create_page_from_content imports with nothing lost (section types, content fields, layout directives, examples). WHEN TO USE: Before drafting or reformatting copy for import, and whenever the user asks how to write, structure, or format content for LandingRabbit. Read-only; takes no arguments.

get_page_format_guide

Lists all pages in the workspace with their IDs, types, and publish status. Each page includes its draft and published metadata (title, description, slug, author, publish date, social image URL, search-engine visibility) and its published collection id — use it to verify state after imports, update_page_metadata, or set_page_publish_state. WHEN TO USE: When the user wants to edit, style, publish, or view an existing page but you don't have the pageId yet. DO NOT USE: When the user just created a page (you already have the pageId from the creation tool response). Only select a page when the user gave a name, slug, URL, or description that clearly matches one result. If the user merely says "build this" with a reference image and no page target, ask whether to create a new page or edit an existing one instead of guessing.

list_workspace_pages

Bulk-imports existing posts or content pages from an external CMS into LandingRabbit. WHEN TO USE: ONLY when a user is migrating from another platform (WordPress, Contentful, Ghost, Webflow, etc.) and already has their content exported — typically a CSV or data dump that you are parsing on their behalf. Each post requires a pre-existing HTML body. This tool: - Accepts up to 50 posts per call (slug, title, and HTML body required per post) - Parses existing HTML, preserving headings, paragraphs, tables, images, and video embeds - Downloads images from their original URLs and re-uploads them to LandingRabbit storage - Creates pages as drafts (default) or publishes immediately - Groups posts under an optional collection slug DO NOT USE: - To write or generate new content — this tool does not create content - When the user asks for "a blog post about X" — that is not a migration, use create_page - When no existing HTML is available to import - For single one-off pages — use create_page or create_page_from_content If the work takes longer than about 20 seconds, the reply carries an operationId instead of the result; fetch it with get_operation_result. Do not repeat a call that returned an operationId.

migrate_blog_content

Replace links across page versions in this workspace. WHEN TO USE: - When you want to bulk replace URLs on your website. - For example, your call-to-actions might point to a URL that you'd like to change (e.g. "/signup" -> "https://app.example.com/signup") WHEN NOT TO USE: - When you want to create a redirect from an old URL to a new one By default this runs as dry-run (apply=false) and returns the versions that would change. When apply=true, the tool writes updates and marks changed published versions as dirty for republish. If the work takes longer than about 20 seconds, the reply carries an operationId instead of the result; fetch it with get_operation_result. Do not repeat a call that returned an operationId.

replace_links