SEO with RankSpot のリモート MCP サーバーとは何ですか?
SEO with RankSpot のリモート MCP サーバーは、https://mcp.rankspot.ai にホストされた Model Context Protocol エンドポイントです。AI アシスタントはローカルに何もインストール・実行することなく接続できます。
SEO with RankSpot MCP Server認証済み注目
@RankSpot
基本情報
SEO with RankSpot MCP Server について
AI SEO, GEO and AI visibility for your website. Track how ChatGPT, Perplexity and Google AI answers mention your brand, research keywords and competitors, find backlink gaps, write and publish SEO articles, and read Google Search Console data from any MCP client.
#seo#ai seo#geo#aeo#ai visibility#generative engine optimization#keyword research#backlinks#search console#content marketing
接続情報
https://mcp.rankspot.aiセットアップ
claude mcp add rankspot --transport http https://mcp.rankspot.aiSEO with RankSpot MCP Server のツール
71List the workspaces this connection can reach, each with its organization, brand, domain and business context. A workspace is one site or brand; the user may belong to several organizations, such as their own and a client's. When the other tools ask for a `workspace_id`, call this first and pick the workspace the user means; ask them if it is unclear. A connection made for one workspace lists just that one. Read-only.
Brand and business context for this workspace: brand name, domain, what the business does, who it sells to, tone of voice, industry, and the country and language it targets. Call this FIRST when deciding what SEO work is worth doing. Every other tool returns data that only means something in context: whether a keyword is relevant, whether a cited page is worth pitching, whether a competitor actually competes. Without this you are guessing at what the business is. `domain` also tells you which citations and backlinks are already your own, but it is stored as the user typed it, so it may be a bare hostname or a full URL — strip the scheme and any trailing slash before comparing it with the bare hostnames every other tool returns. Any field can be null if onboarding was skipped. Read-only.
List SEO keywords tracked in this RankSpot workspace, with search volume, competition, and AI relevance scores. Call this when the user asks which keywords they track, wants to pick keywords to target, or asks about volume or difficulty for keywords they already have. To find content opportunities — keywords worth writing about that have no content yet — use type="new" with sortBy="compositeScore" (the best single priority signal, blending opportunity and AI relevance). Do not call it to discover brand-new keywords for a domain that is not tracked yet. Trial workspaces can only read the first 20 results, so leave offset at 0 unless you know the plan is active.
Add one or more keywords to track in this workspace. Call this when the user wants to start tracking specific search terms. Duplicates are silently skipped. Enrichment (search volume, competition, AI scores) happens on a background job, so the numbers are NOT available immediately after this call — list keywords later to read them. Returns the count of newly inserted keywords.
Return keywords semantically similar to a given seed keyword, for grouping related terms under a single content topic. Call this when planning content and you want the cluster around one keyword. Takes the keyword UUID (from rankspot_list_keywords), not the text. Returns an empty array if the seed has not been semantically indexed yet.
Soft-delete (archive) a keyword. It is excluded from the default list but recoverable via rankspot_unarchive_keyword or by listing with type=archived. Use this to declutter, not to permanently remove data.
Restore a previously archived keyword so it appears in the default list again.
List competitor brands in this workspace. Defaults to the tracked ones. Pass scope="discovered" for brands RankSpot found in AI answers but is not tracking yet, which is what a "should we track this?" decision needs, or scope="all" for both. Each row carries isTracked, autoDiscovered and mentionCount, so you can tell them apart and rank them.
Start tracking a competitor domain. This is also how you act on a track_competitor action: pass the domain from rankspot_list_competitors scope="discovered" and the brand is promoted to tracked, keeping the mentions already attributed to it. Their keywords and backlinks are then discovered on a background sync that runs roughly every 1–2 weeks, so results are NOT available immediately after this call — set that expectation with the user rather than polling. Workspaces have a competitor limit (3 by default) and this fails once that limit is reached, or if the domain is already tracked.
Stop tracking a competitor and remove it from every list. The AI answers already attributed to the brand are kept, so past visibility scores are unchanged, and rankspot_add_competitor on the same domain restores it. Discovered keywords and backlinks are preserved but their competitor link is cleared. Requires confirm: true — only set it after the user has explicitly confirmed removing this specific competitor.
List referring domains for link-gap analysis: sites that link to your competitors but not yet to you, which are your warmest link-building prospects. Each row is one domain with `competitorCount` (how many competitors it links to), `linksToYou`, and `links[]`, one per target with its page (`urlFrom`), anchor, dofollow and spam score; `competitor: null` is the link to you. Default sort is competitorCount, so a domain linking to several competitors comes first: it already writes about this market. type=competitors (default) = unprocessed domains you can pursue; type=mine = domains linking to you. Use competitorId to scope to one competitor. Trial workspaces can only read the first 20 results.
Set a referring domain's status to track which link opportunities you have acted on. Set "processed" once handled, or "new" to requeue it.
Soft-delete (archive) a backlink. Excluded from the default list but recoverable via rankspot_unarchive_backlink or by listing with type=archived.
Restore a previously archived backlink so it appears in the default list again.
List the SEO work planned in this workspace. An action is one thing to do, and `type` says what kind. Call this when the user asks what is planned, in progress or already done. Pass types=["write_article"] when they ask specifically about content, and leave types off to see everything. Status: new (not started), in_progress (running now), processed (finished). Dismissed actions are never returned.
Get a single action by UUID, including its type, status, linked keywords and the ID of the generated article if one exists. Call this to inspect an action before updating it or generating an article for it.
Create a new action. `type` is required and has no default, and it decides which other fields apply: - `write_article` — plan a piece of content, then call POST /v1/actions/:id/generate to write it. Requires no target. - `update_article` — any edit to an existing article, with the instruction in additionalInstructions. Requires `articleId`. - `earn_link` — go after a referring domain your competitors have. Requires `backlinkId`. - `get_cited` — get onto a page AI engines already cite. Requires `citationId`. - `reply_thread` — reply on a cited Reddit or Quora thread. Requires `citationId`. - `record_video` — make a video of your own, since you cannot be added to someone else's. Requires `citationId`. - `track_competitor` — start tracking a brand RankSpot discovered in AI answers. Requires `competitorId`. - `add_prompt` — track a question buyers ask that no existing prompt covers. Requires no target. - `publish_article` — push a finished article to the workspace integrations. Requires `articleId`. - `request_indexing` — ask Google to index a published page it has missed. Requires `articleId`. - `other` — anything you just want written down. Requires no target. For write_article, call rankspot_generate_article afterwards to write it. For update_article put the actual instruction in additionalInstructions ("add internal links", "the pricing section is out of date"). write_article, update_article and add_prompt also take an optional fanoutQueryIds: the searches engines keep running that make the work worth doing. That is provenance, not a target, so pass keywordIds alongside it. For write_article, linking tracked keywords via keywordIds (UUIDs from rankspot_list_keywords) broadens the article's semantic coverage — a good pattern is a strong "new" keyword plus its rankspot_get_keyword_cluster results. Titles must be unique per workspace. Always set `shortDescription`: it is the line under the title in the list and usually the only text the user reads. Put the fuller reasoning in `description`.
Update an action's status, archive it, or edit its title, description, instructions, category or linked keywords. Passing keywordIds replaces the full set of linked keywords. Content can only be edited while the action is "new": once an executor has started, it is locked and editing returns an error. Status can be set on anything that is not currently being generated, so work done outside RankSpot can be closed off here; archived can be set at any time.
Trigger AI article generation. ALWAYS give the user the returned `url` as a link: it opens the article in RankSpot, where they can watch it being researched and written live. The action must be type "write_article" with status "new"; any other type returns an error. Generation is ASYNCHRONOUS and takes about 5-10 minutes. The article is created up front with status "generating", so this returns { success, articleId, slug, url } immediately. Poll rankspot_get_action until its status is "processed" to know it is finished. Respects the plan's article generation limit and fails if it is exceeded.
Permanently delete an action. This is a HARD DELETE and cannot be undone. If an article was generated from it, the article is preserved and only the link is removed. Requires confirm: true, and only set that after the user has explicitly confirmed deleting this specific action.
List AI-generated articles in this workspace. Call this when the user asks what articles exist or their status. The body is omitted here to keep payloads small — use rankspot_get_article to read a single article's body. Two filters matter when planning work: `search` answers "do we already cover this?", and `slug` turns a published URL back into an article. `firstPublishedAt` is when the article went live and stays put on a re-publish, so it is the publication date; `lastPublishedAt` is when the content last reached readers, which is the field to check for staleness — `updatedAt` moves on any write at all. Every row also carries its Google index state: `isIndexed`, `indexCoverageState` (Search Console's own reason when it is not), `indexCheckedAt` and `indexRequestedAt`. Filter with `indexed: false` rather than reading them all back.
Get a single article by UUID with its body as Markdown (`contentMarkdown`). Call this when the user wants to read, review, or edit a specific article's body, and always right before editing it with rankspot_update_article, so the edit starts from the current text.
Edit an article's body and/or metadata (title, description, slug, cover image, category). To edit the body: call rankspot_get_article, edit its `contentMarkdown`, and send the WHOLE article back as `contentMarkdown`. Never send only the changed part or write "rest unchanged": whatever is missing is deleted. You can add, remove, reorder or rewrite sections. Keep lines of HTML (videos and similar) exactly as they are. Links are [text](url); internal links are dofollow and external ones nofollow automatically. Only when the user asks for a dofollow external link, write <a href="https://example.com" rel="dofollow">anchor</a> (rel="nofollow" does the reverse for an internal link). Always pass a `sessionId`: pick one at the start of the conversation and reuse it on every write in that conversation, so the history shows it as one revision. Every write is kept in the revision history (rankspot_list_article_revisions) and can be undone with rankspot_restore_article_revision. Nothing goes live until rankspot_publish_article. The slug must be unique within the workspace.
Push the article as it is now to one of the workspace's connected integrations, exactly like the Publish button in the editor: published for the first time, or updated in place when it is already there. This makes the content public, so only call it when the user asked to publish.
The last 10 revisions of an article, newest first; the first is the article as it is now. Each is one editing session (`source`: editor, api, generation, restore, baseline). `live` marks what was last published. Use it to find the version to go back to with rankspot_restore_article_revision.
Write an older revision's title, description and body back onto the article. The current version stays in the history, so this can be undone the same way. Nothing is published.
Permanently delete an article. This is a HARD DELETE and cannot be undone. Associated action and keyword links are preserved. Requires confirm: true — only set it after the user has explicitly confirmed deleting this specific article.
List "People also ask" questions discovered from search results for tracked keywords. These are real search queries — use them as FAQ sections in articles to capture featured-snippet positions. Call this when planning FAQ content or looking for question-shaped content ideas. They are auto-populated by RankSpot and cannot be created via the API. Filter by status to separate new from already-processed questions.
Update the status of a "People also ask" question. Mark "processed" once addressed (e.g. added as an FAQ), or "new" to reconsider it.
Soft-delete (archive) a "People also ask" question so it no longer appears in the list. Recoverable — this is not a permanent delete.
List categories used to organise actions and articles. Call this when you need a categoryId to assign to an action or article, or when the user asks how their content is organised.
Create a new category for organising actions and articles. Names must be unique within the workspace. A default color is assigned automatically.
Rename a category. The new name must be unique within the workspace.
Permanently delete a category. This is a HARD DELETE and cannot be undone. Actions and articles assigned to it have their category cleared (not deleted). Requires confirm: true — only set it after the user has explicitly confirmed deleting this specific category.
Get clicks, impressions, CTR, and average position from Google Search Console for the connected property. Call this when the user asks how their pages or queries are performing in Google Search. Requires GSC to be connected from the RankSpot dashboard — fails otherwise. Group by up to 3 dimensions. Dates are YYYY-MM-DD; data lags ~2-3 days and covers at most the last 16 months.
Inspect a single URL with Google Search Console and report its index status. Call this when the user asks whether a page is indexed or why it is not. The page is indexed when indexStatusResult.verdict is PASS; otherwise read coverageState for the reason. Requires GSC connected. The URL must belong to the connected property.
Notify Google via the Indexing API that a URL was added/updated (URL_UPDATED) or removed (URL_DELETED). Call this after publishing or updating a page. Requires a GSC connection with the indexing scope where the account is a verified owner. Success means Google received the notification — it does not guarantee indexing; use rankspot_gsc_inspect_url afterwards to check status.
The headline picture of how AI answer engines see this brand over a period: overall visibility, the score per platform (ChatGPT, Perplexity, Google AI Overview, Google AI Mode), share of voice, citation share, and the ten most-mentioned brands with this workspace's own always included. Start here for almost any question about AI visibility, AI search, LLM mentions or competitors in AI answers — it answers "how am I doing", "on which platform" and "versus whom" in one call. startDate and endDate are REQUIRED. To judge a trend, call it twice with two periods of EQUAL LENGTH and compare; unequal periods measure the calendar rather than the brand. A null score means the period had no data, which is not the same as zero — read responsesCounted before quoting any figure, because 100% across two answers is not the claim 100% across two hundred is. Use rankspot_list_ai_responses instead when the user wants the actual wording of an answer.
List the questions RankSpot asks the AI platforms on every run. Call this when the user asks what they are tracking in AI search, or before adding a prompt so you do not duplicate one. Returns the prompts only, with no scores — use rankspot_ai_visibility_summary for how the brand is performing.
Start tracking a question against the AI platforms. Call this when the user wants to monitor how AI answers a specific question about their category. Phrase it as a real user would ask it ("best crm for small business"), not as a keyword. Results are NOT available immediately: prompts run on a daily schedule, so the first answers appear after the next run. Resubmitting a prompt that was archived restores it. Plans cap how many prompts can be tracked at once; at the limit this returns ai-prompt-limit-exceeded, which means archive one rather than retry.
Change the wording of a tracked prompt, keeping its history: it is the same tracked question, reworded. Use this to fix a typo or sharpen the phrasing. For a genuinely different question add a new prompt instead, or the history will span two questions and the trend will be meaningless.
Archive a prompt so it stops being run and frees an allowance slot. The run history is kept, so past scores stay intact — this is why prompts are archived and never deleted. Reversible with rankspot_unarchive_ai_prompt.
Put an archived prompt back into the daily run. It takes an allowance slot again, so it is refused with ai-prompt-limit-exceeded when the plan is full.
The answers the AI platforms actually gave to the tracked prompts, newest first, each with whether it named this brand and which brands it did name. Call this when the user asks what ChatGPT/Perplexity/Google said, or which prompts they are invisible on (mentioned=false). Covers the whole history unless narrowed by date. Runs that failed are never returned, so an absent answer is never a miss. Returns metadata only — call rankspot_get_ai_response with a row id for the answer text itself.
The full text of a single captured answer as markdown, the pages it cited in the order it cited them, the searches the platform ran while composing it, and every brand it named with position and sentiment. Call this to quote or analyse what an engine actually said. Takes the response UUID from rankspot_list_ai_responses.
The pages AI engines cited while answering the tracked prompts, most-cited first. This is the link-building and content worklist for AI search: the sites the engines trust on these questions. Defaults to pages not yet acted on. Use the domain filter (substring, case-insensitive) to check one site, including this workspace's own. Covers the whole history unless narrowed by date, and the citation count is scoped to whatever period is asked for.
Track which citation opportunities have been acted on. Set processed once the page has been dealt with (outreach sent, content written), or back to new to requeue it. This is workflow state only — it does not change what the engines cited or any score.
Archive a cited page so it drops out of the default list. Use it to declutter irrelevant sources, not to remove data: the page keeps counting towards citation share, because archiving is a statement about the workflow and not about what the engines cited. Reversible.
Put a previously archived cited page back on the worklist so it appears in the default list again.
Run a live Google search and get back the top 10 as `items`, the blocks of the results page in page order. Read `type` on each item and branch on it. By default you get the organic results and discussions_and_forums; pass `types` to ask for ai_overview, people_also_ask, video or related_searches as well, but only when you will read them — one ai_overview block can be longer than every organic result together. THIS COSTS THE USER CREDITS, so use it when the answer changes what you do — who already ranks for a query before you propose an article, what a title has to beat — and not to confirm something the RankSpot tools already told you. Location and language come from the workspace, so results match the rest of RankSpot; you cannot and need not set them.
Fetch one public URL and read it. Returns the page as markdown and nothing else. A second provider is tried automatically when the first cannot read the page, and a page that nothing could read is an error rather than an empty string. THIS COSTS THE USER CREDITS per page. Use it to settle a specific question a listing cannot answer: does this directory accept submissions, does this roundup already mention us, is this thread still active. Never crawl a site page by page — if you want several pages from one domain, fetch the one that decides the answer.
The searches AI engines ran while answering the tracked prompts, most-run first. Where a citation says which page an engine trusted, a fanout query says what it decided it needed to know — so a query that comes up often and has no content behind it is a gap stated in the engine's own words, and one of the strongest signals here for what to write next. Feed a promising one into rankspot_create_action as fanoutQueryIds. State is derived from the work linked to the query rather than stored, so it moves on its own as actions and articles appear. Defaults to queries with nothing planned or written for them. Covers the whole history unless narrowed by date; a date range lists only the queries engines ran inside it, and scopes the search count to it as well.
The searches that mean the same thing as this one, most-run first, with the seed leading. Engines phrase one intent many ways, so an action should carry the cluster rather than whichever phrasing you happened to pick: pass the ids straight to rankspot_create_action as fanoutQueryIds. Takes the search UUID from rankspot_list_ai_fanout_queries. Returns just the seed when nothing else is close.
Archive a fanout query so it drops out of every non-archived list. This is the only way to retire a search by hand, since the rest of its state follows the actions and articles linked to it. Use it to declutter navigational or irrelevant searches, not to remove data: it keeps counting towards the reporting, because archiving is a statement about the workflow and not about what the engines ran. Reversible.
Put a previously archived fanout query back on the worklist so it appears in the default list again.
Reads a page, works out who wrote it, and finds and checks their email. Returns the author and up to 5 addresses best first, each with `deliverability` (valid, risky = the domain accepts everything, unknown, not_checked) and `confidence`. Email the first; pass the rest as backups in `contactEmails`. THIS COSTS THE USER CREDITS (about 10 to 40 per page) and can take a minute. Before paying for a lookup, check the site was not already contacted with rankspot_list_outreach and `domain`. If you already know the person, rankspot_find_email is cheaper.
Finds and checks the address of a known person at a domain. `email` is null when nothing deliverable was found. COSTS CREDITS, fewer than rankspot_find_contact since no page is read.
Outreach conversations, newest activity first. Use it to check whether a site or person was already contacted (`domain`, `email`) before finding contacts or pitching, and to see what is new: `needsReply: true` lists replies waiting on the user, `since` anything that moved after a time. `status`: scheduled (nothing sent yet, still editable), sent (no reply yet), replied, processed. `pending: true` lists everything with an email still to go out, soonest first. Read a thread with rankspot_get_outreach.
One opportunity: the emails as written (`sequence`, each with its status and not-before time) and everything exchanged so far (`thread`, oldest first; a reply's `body` has the quoted earlier emails removed).
Creates an outreach opportunity with emails YOU write, and schedules them to go out from the workspace's mailboxes. Show the user the emails and get their go-ahead before calling: these are real emails to real people. The first waits at least an hour (it can be edited or deleted until it is sent), then goes out at the next free slot: 10 a day per mailbox, every day 8:00 to 20:00 US Central. Follow-ups go 3 and 7 days later unless someone replies. Rules, checked on every email: no links or domain names anywhere; greet with {{first_name}} rather than a typed name; the first email must sign off with {{sender_name}} or {{sender_first_name}} (the mailbox is picked when it sends). A 400 lists every problem at once. Refuses an address already contacted, and a site already contacted unless `allowContactedSite` is true (another author at the same publisher), and 409s with `outreach-no-mailbox` until a mailbox is connected in the RankSpot dashboard. Check capacity with rankspot_list_mailboxes before queueing a lot.
Changes any of the fields until the first email is sent (status scheduled). `sequence` replaces the whole list. The send time stays as it was. Refused (409) once the first email went out.
Permanently deletes an opportunity nothing was sent from yet. Once a conversation started, use rankspot_stop_outreach instead.
Cancels every scheduled email of a conversation. The thread and any replies stay.
Sends a reply right away from the conversation's mailbox to whoever wrote last, threaded under the exchange, and cancels follow-ups still waiting. It is a real email: show the user what you will send and get their go-ahead first. Plain text; links are fine here, unlike in the automated emails.
Marks it processed, whatever came of it: anything waiting is cancelled and it becomes read-only. A late reply is still recorded.
Outreach numbers for a period: emails sent (first emails and each follow-up), replies from people, automatic replies, bounces, reply rate (of the sites first emailed in the period, how many answered), bounce rate, which email in the sequence the replies answered, median hours to a reply, and per-day counts. Use it to answer "how is outreach going" or to compare periods. Days are US Central; the default is the last 30 days.
Each connected mailbox with what it sent today and how many emails are bound to it (`queued`, mostly follow-ups; a first email is bound once a mailbox picks it). Each mailbox sends 10 automated emails per day, so check this before queueing a lot of outreach. Mailboxes are connected in the RankSpot dashboard, not here.
The paid, exchanged and free links logged for this workspace, newest first. Use `status: pending` to see who still owes a link. Statuses are set by hand: nothing checks that a link is really live.
Records a paid, exchanged or free link. Only `type` is required, and it starts `pending` until their link is up. For an exchange, `url` is their page linking to us and `ourUrl` our page linking to them. Pass `outreachId` when the deal came from a conversation.
Changes any of the fields, such as `status: live` once their link is up or `status: lost` when they dropped it. Null clears a field, except `type`, `status` and `dofollow`.
Permanently deletes the record. To remember a link they dropped, set `status: lost` with rankspot_update_outreach_link instead.
概要
What is SEO with RankSpot?
SEO with RankSpot is the official remote MCP server for RankSpot, an AI SEO and AI visibility platform. It connects Claude, ChatGPT, Cursor, Codex, Gemini CLI and any other MCP client to your RankSpot workspaces, so your agent can do SEO research, planning and writing in plain English.
It is hosted, so there is nothing to install. Point your client at https://mcp.rankspot.ai and sign in once with OAuth.
How to use SEO with RankSpot?
- Create a RankSpot account at rankspot.ai and set up a workspace for your site.
- Add
https://mcp.rankspot.aito your MCP client as a remote server (Streamable HTTP). - On the first tool call, your client opens a RankSpot sign-in page. Log in, review the workspaces the agent will reach, and approve.
- Ask in plain English, for example: "How often is my brand mentioned in AI answers this month?"
Prefer an API key? Create one under Account settings → API Keys and send it as Authorization: Bearer YOUR_RANKSPOT_API_KEY.
Key features of SEO with RankSpot
- AI visibility (GEO / AEO): see how ChatGPT, Perplexity and Google AI answers describe your brand, which pages they cite, and which searches they run.
- Keyword research: track keywords, read search volume and competition, and pull the semantic cluster around any seed term.
- Competitor and backlink analysis: track competitor domains and find sites linking to them but not to you.
- Content planning and writing: plan SEO work as actions, generate publish-ready articles, edit them, and publish to your connected site.
- Google Search Console: read clicks, impressions, CTR and position, check whether a page is indexed, and request indexing.
- Live research: run a live Google search or fetch any public page as clean markdown.
- Link-building outreach: find the right contact on a site and send outreach from your connected mailbox, then track replies.
- Safe by default: every tool is annotated read-only or destructive, and deletes require explicit confirmation.
- Pay as you go: no subscription. Reading your data is free; paid actions like articles and outreach are charged per action, with the price shown first. Start with $1 free credit, top up from $10.
Use cases of SEO with RankSpot
- Track your brand's share of voice in AI search answers and see which competitors show up more.
- Find Search Console queries stuck on page two and plan content to push them up.
- Spot content gaps against competitors and turn each one into a planned article.
- Generate and publish SEO articles without leaving your agent.
- Find backlink opportunities your competitors have and you don't.
FAQ from SEO with RankSpot
Do I need a RankSpot account? Yes. You need a RankSpot account with at least one workspace for your site.
Which clients does it work with? Any MCP client that supports remote servers over Streamable HTTP, including Claude, Claude Code, ChatGPT, Cursor, Codex, VS Code, Gemini CLI and Antigravity.
How do I authenticate? OAuth is recommended: one browser sign-in, tokens refresh automatically, and you can revoke access under Account settings → Connected Apps. An API key also works for clients without OAuth.
What do I need connected? Search Console tools need a connected Search Console property, publishing needs a connected website, and outreach needs a connected mailbox.
How much does it cost? RankSpot is pay as you go, with no subscription. Start free: sign up, connect your agent and read your data at no cost, with $1 of credit to try paid actions. RankSpot charges per action (per article, per email found, per email sent), and every price is shown before it runs. Top up from $10 when you want more; credit rolls over and nothing renews. See pricing.
よくある質問
SEO with RankSpot の MCP サーバーにはどう接続しますか?
エンドポイント https://mcp.rankspot.ai を Claude Code、Cursor、VS Code など任意の MCP 対応クライアントに追加してください。このページのセットアップ手順で各クライアントを一度に設定できます。
SEO with RankSpot の MCP サーバーは認証が必要ですか?
はい。SEO with RankSpot は OAuth を使用します。初回接続時に MCP クライアントがブラウザを開いてサインイン・認可を行い、以降はその認証情報を再利用します。
SEO with RankSpot の MCP サーバーはどのトランスポートを使用しますか?
SEO with RankSpot は Streamable HTTP エンドポイントを公開しており、これはリモート MCP サーバーで広く使われ、主要な MCP クライアントがサポートするトランスポートです。
他のリモート MCP サーバー
2pr.io
Write, schedule and publish LinkedIn posts in your own voice, reply to comments and read your analytics, all from your AI assistant. Remote MCP server with OAuth sign-in.
A
AI-Ready Webpage Extractor (Apify)
Convert web pages to clean Markdown and retrieval-ready chunks with browser rendering support.
A
Aayat AI
Pay-per-use data and tools for AI agents: crypto token safety and rug checks, web search with cited answers, page-to-Markdown, up-to-date library docs, package safety, company research and UK/EU open data. 160 tools, 20
A
AccountHub
One MCP connection for all your Gmail, Google Calendar, Drive and Contacts accounts, Slack workspaces and Notion — free.
AOI — Environmental Intelligence
AOI is a commercial environmental-intelligence gateway that gives AI agents programmatic access to 38 real-world intelligence capabilities — climate risk, satellite monitoring, methane detection, water & soil, vegetation
コメント