Remote server (HTTP + OAuth)
Nothing to install. You paste a URL, sign in once, and the connector works on the web, desktop and mobile. This is how mcp.speedyindex.com/mcp works.
MCP server · Claude connector
SEO MCP server for Claude, Cursor and any AI assistant
Check indexing in Google, Bing and Yandex, submit URLs with pay-per-indexed refunds, verify backlinks and track rankings — in plain words. 22 tools, one URL.
https://mcp.speedyindex.com/mcpPaste it as a custom connector in Claude, Cursor or VS Code and sign in with your SpeedyIndex API key. New accounts get 100 free tokens.
Which of my 5 new blog URLs are not in Google yet? Send those for indexing.
tool create_index_check_task engine: google · 5 URLs
tool get_task_report
3 of 5 are indexed. Not indexed yet:
Shall I send the 2 missing pages for indexing? URLs that don’t get indexed are refunded.
| /blog/mcp-for-seo | not indexed |
| /blog/claude-connectors | not indexed |
Yes, go.
tool create_indexing_task engine: google · 2 URLs
Done — task created. I’ll check the report when it’s ready.
What is an MCP server — and why SEOs need one
The Model Context Protocol (MCP) is an open standard, introduced by Anthropic in November 2024, that lets AI assistants call external tools. An MCP server exposes a service’s actions as tools; a connector is what Claude calls an MCP server you have added. Instead of exporting CSV files and pasting them into the chat, the assistant reads live data and runs tasks for you — and shows you every call it makes.
Your AI assistantClaude, Claude Code, Cursor, VS Code or another MCP client
⇄MCP
SpeedyIndex MCP servermcp.speedyindex.com — 22 tools, OAuth sign-in
⇄API
Your SpeedyIndex accountIndexing, index checks, backlinks, Rank Tracker
Local server (stdio)
Runs on your computer through npx, with the API key in an environment variable. Useful for clients that read a local config file. Package: speedyindex-mcp.
Tools, not answers
The assistant decides which tool to call, but every result comes from the service itself — real index status, real positions — not from the model’s memory.
MCP server vs connector vs plugin
An MCP server is the service that exposes tools. A connector is what Claude calls an MCP server you’ve added to your account. A plugin or extension is a client-specific add-on that may bundle MCP servers, prompts and settings. SpeedyIndex is one MCP server: add it as a connector in Claude, or as a server in Cursor and VS Code.
Why SEOs connect MCP servers to Claude and Cursor
Live data, not guesses
A model alone can’t know whether your page is indexed today. With a connector it checks.
Chains of tasks in one prompt
Check → send the misses for indexing → re-check later. One request instead of three dashboards.
Reports in plain language
Ask for a client summary and get tables and conclusions, not raw exports.
Works where you work
Same tools in the Claude chat, in Claude Code next to your site’s code, and in Cursor or VS Code.
SEO MCP servers by category
No single server covers all of SEO. Most teams combine a data source for keywords and backlinks, first-party data from Google, and task tools. SpeedyIndex fills the gap the popular lists leave open: getting pages indexed and proving they are indexed, plus rank tracking.
| Category | Examples | What you use it for | How you pay |
|---|---|---|---|
| Keyword and backlink databases | Ahrefs, Semrush, SE Ranking, DataForSEO | Keyword volumes and difficulty, competitor research, backlink profiles from the vendor’s index | Your plan with the vendor; DataForSEO charges per data request |
| First-party Google data | Search Console (community-built servers); GA4 (Google’s experimental server) | Clicks and impressions, URL inspection one page at a time (about 2,000 inspections a day per property), sitemap submission, traffic | Free, with access to your property |
| Crawling | Screaming Frog, Firecrawl | Site audits, page content and technical checks | Licence (Screaming Frog, runs locally) or credits (Firecrawl) |
| Indexing, index checks, rank tracking | SpeedyIndex | Index status of up to 10,000 URLs per task in Google, Yandex and Bing; submitting URLs for indexing; checking that backlinks are live; Google positions and AI Overview citations | Tokens, no subscription; Google indexing is pay-per-indexed |
The categories reflect what each server is mainly used for; check each vendor’s page for current tools and prices. Google does not publish an official Search Console MCP server — the servers listed in MCP directories are community projects.
A workflow that uses them together
- Plan: pick keywords with your keyword database connector.
- Publish the page, then ask SpeedyIndex to check whether it is in Google — and send it for indexing if not.
- Track: add the keywords to a Rank Tracker project and see positions and AI Overview citations after each check.
- Build links and let the backlink checker confirm that each donor still links to you and that the donor page is itself indexed.
- Report: ask for a monthly summary across all of it.
What the SpeedyIndex connector can do: 22 tools
15 tools only read data and are free. 7 change your account, so your client asks before running them (unless you’ve set a tool to “Always allow”). 5 of those, marked in yellow, spend tokens or credits.
Indexing in Google and Yandex
Send up to 10,000 URLs per task. Google tasks are pay-per-indexed: tokens for URLs that don’t get indexed come back automatically. Spread a task over 2–30 days with drip feed.
create_indexing_tasksubmit_single_urllist_tasksget_task_statusget_task_reportIndex checks: Google, Yandex, Bing
Find out which URLs are actually in the index — up to 10,000 per task — and get indexed and not-indexed lists (with page titles for Google).
create_index_check_taskget_task_reportget_balanceBacklink checker
Open donor pages and check that they still link to your domains: HTTP status, anchor text, nofollow flag, Moz Domain Authority and spam score for each donor.
create_backlink_check_taskget_backlink_check_statusget_backlink_check_reportGoogle Rank Tracker
Create projects, add keywords and competitors, run checks and read positions, changes since the last check, trend history and whether Google’s AI Overview cites your site.
rank_tracker_start_checkrank_tracker_create_projectrank_tracker_add_keywordsrank_tracker_get_positionsrank_tracker_get_trendrank_tracker_get_keyword_detailsrank_tracker_get_check_cost+5 moreyellow = spends tokens or credits. Creating a Rank Tracker project or adding keywords is free; running a check is not.
How to connect the SpeedyIndex MCP server
You need a SpeedyIndex account and its API key: log in at app.speedyindex.com and copy the key from the API section (it is also in the @speedyindexbot settings). New accounts get 100 free tokens. The connector itself is free — you pay only for tasks you run.
- Open Claude → Customize → Connectors and choose Add custom connector.
- Name it SpeedyIndex and paste the server URL
https://mcp.speedyindex.com/mcp. Select Add. - Select Connect. A SpeedyIndex sign-in page opens: paste your API key and confirm. You return to Claude, signed in.
- In a chat, click + in the message box → Connectors and turn SpeedyIndex on. Ask: “What is my SpeedyIndex balance?” — if you see tokens and credits, you’re set.
- Custom connectors work on every Claude plan (the Free plan allows one). On Team and Enterprise, an owner adds connectors in the organization settings.
Add the hosted server (with --scope user it works in every project):
claude mcp add --transport http --scope user speedyindex https://mcp.speedyindex.com/mcp
Then run /mcp inside Claude Code, choose speedyindex → Authenticate and sign in in the browser. claude mcp list should show it as connected.
Prefer a local server with your key in an environment variable? Use the npm package:
claude mcp add speedyindex -e SPEEDYINDEX_API_KEY=your-key -- npx -y speedyindex-mcp
Add this to .cursor/mcp.json (project) or ~/.cursor/mcp.json (all projects). Cursor signs you in with OAuth when the server first connects; turn the server on under Customize. Check it by asking for your SpeedyIndex balance and expanding the tool call.
{
"mcpServers": {
"speedyindex": { "url": "https://mcp.speedyindex.com/mcp" }
}
}Any client that supports remote MCP servers over Streamable HTTP can use the server URL. In VS Code, run MCP: Add Server → HTTP and paste the URL, or add it to a workspace .mcp.json:
{
"mcpServers": {
"speedyindex": { "type": "http", "url": "https://mcp.speedyindex.com/mcp" }
}
}Check it with MCP: List Servers, then ask for your balance — a plausible answer without a tool call doesn’t prove the connection.
Clients that only run local (stdio) servers — for example Claude Desktop’s config file — can use the npm package:
{
"mcpServers": {
"speedyindex": {
"command": "npx",
"args": ["-y", "speedyindex-mcp"],
"env": { "SPEEDYINDEX_API_KEY": "your-key" }
}
}
}If sign-in fails
- Paste the API key without spaces; copy it again from the dashboard if SpeedyIndex rejects it.
- Leave the OAuth Client ID and Secret fields empty — the server registers clients automatically.
- The sign-in page is valid for 10 minutes. If it expired, remove the connector and add it again.
- On Claude Team or Enterprise, ask an owner to add the connector for the organization.
- Cursor: check Output → MCP Logs. Claude Code: run
/mcpand choose Authenticate again. - Still stuck? Write to support@speedyindex.com.
12 ready-to-use prompts for SEO workflows
Type these in plain words. The assistant picks the right tools and, with default settings, asks before it runs anything paid.
Index check
Check which of these 40 URLs are indexed in Google and list the ones that are not.Uses create_index_check_task, then get_task_report.
Indexing
Send the unindexed URLs from that check to Google indexing and give me the task id.Pay-per-indexed: no charge for URLs that don’t get indexed.
New content
Here are 300 new URLs (pasted below). Drip-feed them to Google over 10 days.create_indexing_task with drip_feed_days: 10.
Bing & Yandex
Are these product pages in Bing and Yandex? Make a table by engine.Index checks in both engines from one prompt.
Backlinks
Which of these 25 donor pages still link to example.com, and which links are nofollow?Backlink checker with status, anchor, DA and spam score.
Link audit
From these 50 guest-post URLs, find the ones that lost my link and the ones that aren’t indexed — I want a re-indexing list.Combines backlink check, index check and indexing.
Rank tracking
Create a Rank Tracker project for example.com in the United States with these 30 keywords and 3 competitors.Creating the project is free; ask the cost before the first check.
Ranking drops
Which keywords dropped since the last check, and what page ranks for them now?rank_tracker_get_positions with changed_only: down.
AI Overviews
In my example.com Rank Tracker project, does the AI Overview for “best running shoes” cite my site? Who does it cite?rank_tracker_get_keyword_details shows AI Overview sources.
Client report
Summarize this month for my client: visibility trend, top-10 keywords gained and lost, and indexing results.Trend + positions + task reports in one answer.
Yandex
Send these 50 Russian-language pages to Yandex indexing.create_indexing_task with engine: yandex.
Migration
We just migrated to a new domain. Check which of these 500 new URLs are indexed in Google and send the missing ones for indexing over 7 days.Index check, then drip-feed indexing. See the post-migration checklist.
Security and costs
Your key stays private
You enter the API key on our sign-in page, not in the chat. The server has no database: the key lives only inside an encrypted sign-in token (AES-256-GCM). Access tokens last 1 hour; sign-ins renew for up to 90 days. Regenerate the key in the dashboard to cut off every sign-in at once.
Paid actions need your OK
Tools that spend tokens are marked for the client. Claude asks for your approval before running them unless you’ve set that tool to “Always allow”; other clients follow their own approval settings. The connector can’t pay, create invoices or delete your projects and tasks.
Pay only for results
Google indexing is pay-per-indexed: URLs that don’t get indexed are refunded automatically. Rank Tracker checks hold the maximum cost and return the unused part when the check ends.
Details: connector privacy policy · technical docs · SpeedyIndex API
Where to find the SpeedyIndex MCP server
The same server is listed in the main MCP directories. Pick the one your client reads from, or star it on GitHub.
PDF guideMCP for SEOSpeedyIndex
Free PDF: MCP for SEO — a practical guide
What MCP is, how to connect it in Claude, Claude Code, Cursor and VS Code, and 20 copy-ready prompts for indexing, index checks, backlinks and rank tracking. 12 pages, no sign-up.
Frequently asked questions
What is an SEO MCP server?
It is an MCP server that gives an AI assistant SEO tools: for example checking index status, submitting URLs for indexing, reading rankings or backlink data. The SpeedyIndex MCP server has 22 such tools for indexing, index checks in Google, Yandex and Bing, backlink checks and Google rank tracking.
Which MCP server is best for SEO?
There is no single best one. Most teams combine a keyword and backlink database (Ahrefs, Semrush, SE Ranking or DataForSEO), Search Console data, and a task server such as SpeedyIndex for index checks, indexing and rank tracking.
Is there a free SEO MCP server?
The SpeedyIndex connector is free to add, and its read tools cost nothing. Paid tasks use tokens, and new accounts get 100 free tokens. Community Search Console servers are free with access to your property.
How do I add the SpeedyIndex connector to Claude?
In Claude open Customize → Connectors, choose Add custom connector and paste https://mcp.speedyindex.com/mcp. Select Connect, enter your SpeedyIndex API key on the sign-in page and confirm. Custom connectors are available on all Claude plans; the Free plan allows one.
Do I need to be a developer?
No. In Claude you paste one URL and sign in. Config files are only needed for Cursor, VS Code or a local npm setup, and the snippets above can be copied as they are.
Is the connector free?
Yes, the connector itself is free. You pay with SpeedyIndex tokens or credits only for tasks that cost money: index checks, indexing, backlink checks and Rank Tracker checks. New accounts get 100 free tokens, and Google indexing is pay-per-indexed.
Can the AI spend my balance without asking?
Tools that spend tokens are marked for the client, and Claude asks for your approval before running them unless you have set that tool to “Always allow”. Other clients follow their own approval settings. Read-only tools like balance, reports and positions cost nothing. The connector cannot pay, create invoices or delete projects and tasks.
Where is my API key stored?
You enter the key on the SpeedyIndex sign-in page, not in the chat. The server has no database: the key is kept only inside an encrypted sign-in token held by your client. Regenerate the key in the dashboard to revoke every sign-in.
Can Claude check whether my pages are indexed in Google?
Yes. Ask it to check a list of URLs; it creates an index check task for up to 10,000 URLs and returns indexed and not-indexed lists. The same works for Yandex and Bing.
Can Claude submit URLs to Google for indexing?
Yes, through SpeedyIndex. Google tasks are pay-per-indexed: tokens for URLs that don’t get indexed are refunded automatically. You can also spread a task over 2 to 30 days with drip feed.
Can I use several SEO MCP servers at once?
Yes. Clients like Claude, Cursor and VS Code can have several servers connected; the assistant picks tools from each. A common setup is a keyword database connector plus SpeedyIndex for indexing and rank tracking.
Which clients does it work with?
Claude (web, desktop and mobile), Claude Code, Cursor, VS Code and any client that supports remote MCP servers over Streamable HTTP. Clients that only run local servers can use the npm package speedyindex-mcp.