Citebase: AI SEO & AI Visibility Toolkit (MCP Server)

Connect Citebase to your AI assistant (MCP setup and sign-in)

Citebase is an MCP server. Add it once, sign in once, then ask for help in plain words.

You need: an Apify account and an assistant that supports MCP connectors (Claude, ChatGPT, Cursor and others). Create an Apify account at apify.com if you do not have one. The account you sign in with pays the small run cost and any data jobs you approve.

Setup address (use this everywhere):

https://mcp.apify.com/?actors=khadinakbar/citebase&tools=fetch-actor-details,call-actor,get-actor-run,get-dataset-items,get-key-value-store-record,abort-actor-run

The extra tools in this address let your assistant run the paid data jobs you approve. On the web nothing but your assistant's own confirmation prompt stops a run, so approve only the Actor and the cost it quoted. If your assistant cannot run data jobs, add the connector again using the setup address.

Two ways to sign in

Way	Use it when	What you do
Apify sign-in (OAuth), recommended	Your assistant has a "Connect" or "Sign in" button. ChatGPT only supports this way.	Add the setup address, press Connect, sign in to Apify, press Allow. No token to copy.
Apify API token	Your assistant or script cannot do OAuth	Send the header Authorization: Bearer YOUR_APIFY_TOKEN. Find the token in Apify Console under Settings, then API and integrations.

Steps for each assistant (menu names can differ a little by plan and app version)

• Claude (claude.ai and Claude Desktop): open Settings, then Connectors, then Add custom connector. Name it Citebase, paste the setup address, press Connect and sign in to Apify.

• ChatGPT: open Settings, then Connectors, turn on Developer mode, create a new connector, paste the setup address, choose OAuth, then sign in to Apify.

• Claude Code: run this in your terminal, then type /mcp inside Claude Code, pick citebase and sign in.

  $claude mcp add --transport http citebase "https://mcp.apify.com/?actors=khadinakbar/citebase&tools=fetch-actor-details,call-actor,get-actor-run,get-dataset-items,get-key-value-store-record,abort-actor-run"
  

  To use a token instead, add --header "Authorization: Bearer YOUR_APIFY_TOKEN" to that command.

• Cursor and other apps that use a JSON settings file: add this block. Leave out headers to sign in with OAuth when the app supports it. Field names can differ slightly by app.

  {
    "mcpServers": {
      "citebase": {
        "url": "https://mcp.apify.com/?actors=khadinakbar/citebase&tools=fetch-actor-details,call-actor,get-actor-run,get-dataset-items,get-key-value-store-record,abort-actor-run",
        "headers": { "Authorization": "Bearer YOUR_APIFY_TOKEN" }
      }
    }
  }
  

• Direct connection (advanced): use this Actor's own Standby address (the one ending in /mcp, shown on this Actor's API tab in Apify Console). It needs the token header and does not offer a sign-in button.

Check that it works. Paste your website address and ask for a quick health check. Your assistant should come back with a score in words and a short list of fixes. Citebase adds no fee for this; Apify's small run cost applies.

Keep your token safe.

• Prefer the sign-in route. A token gives access to your whole Apify account.
• Never type a token into a chat or into a tool field. Citebase never asks for passwords or keys.
• If a token leaks, delete it in Apify Console and make a new one.

If sign-in fails, see the troubleshooting table near the end of this page.

What is Citebase?

Citebase is a connector that helps a brand get recommended and cited by AI assistants such as ChatGPT, Claude, Gemini and Perplexity. You connect it to your own AI assistant and ask for help in plain words. It returns step-by-step guidelines, a website health check, price quotes for paid data jobs, "how often does AI recommend me" results, a short to-do list, a check for outreach emails, saved history, a shareable report and a plan for automatic re-checks. Citebase itself adds no fee; the data jobs cost money and come from separate Actors listed below, always with a price quote and a spending cap first.

Built for founders, marketers and small agencies who want to show up when people ask an AI "what is the best X for Y?" and do not want to learn SEO jargon first.

Choose Citebase when / do not use it when

Choose Citebase when you want to:

• check a website for search and AI-readiness problems, with a score and a plain list of fixes;
• find out whether AI assistants recommend your brand, repeat the test later and see whether your changes helped;
• plan links, press and outreach with checklists, and check an email draft before it is sent;
• know what to do next, in plain words, ranked, with who does it and how long it takes;
• work from one assistant (ChatGPT, Claude, Cursor or any assistant that supports MCP connectors) without five separate SEO tools.

Do not use it to:

• scrape websites or collect data yourself (use the paid Actors listed below, through Citebase's quotes);
• send email (it only checks drafts and never sends);
• read private, local or password-protected pages (the site check refuses them);
• get live Google rankings by itself (that is the serp_page job).

How a session works: from a website address to a plan

1. Start. You paste your website address. The assistant reads the start playbook and checks your site first; it asks only what it still needs.
2. Check the site. audit_site scores the site and lists the top problems.
3. Price the research. quote_cost totals the paid data jobs (search results, keywords, backlinks, AI answers) and gives a spending cap.
4. Research. With your yes, the assistant runs the Actors below with that cap.
5. Test the AI. It builds a fixed list of questions customers might ask, collects the AI answers several times and sends them to measure_recommendation_share.
6. Plan. next_steps turns the audit and results into a ranked to-do list.
7. Reach out. For links and press, the assistant drafts emails and runs check_outreach_draft before asking you to approve each one.
8. Re-check. After about 30 days it repeats the test to see whether the changes worked.

Tools

Tool	What it returns	Price
get_playbook	Citebase's guidelines: the list of names, or one playbook's text (22 workflows and 12 guides and checklists)	No Citebase fee
audit_site	0-100 health score, scores per area, the top problems with example pages	No Citebase fee
quote_cost	Cost of each paid job, the total and the spending cap to use	No Citebase fee
measure_recommendation_share	How often AI recommends the brand, with ranges, a noise check and a verdict per change	No Citebase fee
next_steps	Ranked to-do list: what, why, who, how long	No Citebase fee
check_outreach_draft	Ready yes/no, problems to fix, warnings, a fingerprint of the exact text	No Citebase fee
check_ai_content	AI-writing pattern density for a draft (local style lint; not a commercial detector)	No Citebase fee
save_results	Saves AI-answer rounds and site checks under a project name in your own Apify account	No Citebase fee
get_history	Recommendation Share and site score over time, plus logged changes	No Citebase fee
make_report	A one-page report with a trend chart and a link you can share	No Citebase fee
import_ai_results	Converts results of AI Brand Monitor, Google AI Overviews, Google AI Mode and the tracker Actors into saved rows; a first round goes in with one call (send the questions and brand with it); flags close calls and rounds that were asked under different conditions	No Citebase fee
schedule_checks	Cron text, the Actor runs to add (assistants repeated 1 to 5 times, Google answers once) and the worst-case monthly cost; the website comes from the saved project	No Citebase fee

What one result contains

audit_site returns one object:

Field	Type	Meaning	Example
site	text	The site that was checked	https://example.com
pages_checked	number	Pages read	3
health_score	number or null	0-100, higher is better	66
subscores	object	0-100 per area	{"technical": 70, "on-page": 55}
counts	object	Checks failed, warned, passed	{"fail": 2, "warn": 5, "pass": 20}
problems	list	Up to 25, worst first: level, area, check, detail, example pages	see below
passed	list	Names of checks that passed	["HTTP redirects to HTTPS"]
limits	text	What the check cannot see	"no JavaScript is run"

{
  "ok": true,
  "tool": "quote_cost",
  "result": {
    "lines": [{"job": "serp_page", "actor": "khadinakbar/scrape-google-serp", "units": 15, "unit": "SERP page", "requests": 1, "cost_usd": 0.0721, "compute_usd": 0.045}],
    "total_usd": 0.0721,
    "suggested_cap_usd": 0.1,
    "includes_compute_usd": 0.045,
    "compute_note": "This total includes about $0.05 of Apify compute, which Apify bills on top of the data prices.",
    "prices_read_on": "2026-11-08"
  }
}

Every tool has a typed output schema, so an assistant can read the exact shape before calling.

Quick start and input

1. Connect your assistant using the steps at the top of this page.
2. Paste your website address and ask for a quick health check.

Plain JSON call, POST /run (what an assistant sends through mcp.apify.com):

{"tool": "audit_site", "arguments": {"url": "https://example.com", "max_pages": 10}}

You can also run this Actor normally with the same two fields as input: tool (one of the twelve names, or scheduled_check: the site check plus report refresh, meant for an Apify Schedule) and arguments (a JSON object). A bad input ends the run as failed and the dataset item says what to fix.

Actors Citebase uses for paid data

Citebase never runs these itself and never calls other people's Actors. It tells your assistant which one fits, what it will cost and which spending cap to set; you approve and the assistant runs it on your own Apify account. All are public Actors by the same author.

Job name	Actor	What it gets you	Price
site_audit	Complete SEO Audit	Technical site audit: crawl rules, redirects, headings, links, structured data (server HTML, no JavaScript)	$0.10 per audit
serp_page	Google SERP Scraper	Google search results pages: who ranks, questions people ask, search features	$0.0018 per SERP page
keyword_ideas	Keyword Research	Keyword ideas with search volume and difficulty	$0.20 per run + $0.01 per keyword row (cannot run until 2026-11-14)
keyword_volume	Keyword Research	Exact search volume, cost per click and competition for keywords you already have	$1.00 per run + $0.01 per keyword row (cannot run until 2026-11-14)
backlinks	Website Backlink Checker	Who links to a site: backlinks, referring domains and anchor text	$0.05 per run + $0.01 per backlink or domain row
backlink_prospects	Backlink Opportunity Finder	Link prospects: resource pages, best-of lists and guest-post sites	$0.006 per prospect
ai_brand_check	AI Brand Monitor	Does an AI assistant mention or recommend your brand? Checks ChatGPT, Perplexity, Claude and Gemini answers	$0.08 per prompt x engine x run
ai_page_citation	AI Search Visibility Tracker	Which of your pages AI assistants cite, and which competitors they cite instead	$0.09 per keyword x query x engine
ai_citation_scan	AI Citations Tracker	The sources behind a Perplexity answer	$0.12 per Perplexity prompt
ai_overview	Google AI Overviews Scraper	Google AI Overviews for a search, and whether your domain is among the sources	$0.015 per query, $0.10 when the fallback route is used (it was in every test run)
ai_mode	Google AI Mode Scraper	Google AI Mode answers and their citations	$0.025 per query
website_contact	Bulk Website Contact Extractor	Public emails and social links from a website (for outreach)	$0.005 per domain
email_verify	Email Address Validator	Checks that an email address is real before you send	$0.0009 per email
news_article	Google News Scraper	News coverage and the journalists who wrote it (for PR)	$0.003 per article
reddit_result	Reddit Posts and Comments Scraper	Reddit posts and comments on a topic	$0.003 per result
youtube_result	YouTube Search Scraper	YouTube videos and channels on a topic	$0.003 per result
review_site	Software Reviews Scraper	Software reviews from G2, Capterra and TrustRadius	$0.007 per review, plus $0.01 per product found and a start fee of $0.005 per GB of memory
trustpilot_review	Trustpilot Reviews Scraper	Trustpilot ratings, reviews and replies	$0.0025 per review
google_trend	Google Trends	Interest over time, regions and rising searches for a topic	$0.005 per trend item
search_suggestion	Google Autocomplete Suggestions	What people start typing in Google for a topic	$0.005 per lookup
longtail_keyword	Long-Tail Keyword Ideas	Many specific long-tail phrases grown from a seed word	$0.0015 per keyword
keyword_difficulty	Keyword Difficulty Checker	How hard each keyword is to rank for, in batches	$0.30 per batch
keyword_metrics	Keyword Metrics	Search numbers, matching terms and related terms for keywords	$0.01 per result
content_brief	SEO Content Brief Generator	A page outline built from what the top results cover	$0.02 per run + $0.005 per brief
quora_question	Quora Scraper	Real questions and answers people post on a topic	$0.003 per item, plus a start fee of $0.005 per GB of memory
reddit_search	Reddit Search Scraper	Reddit posts that match a search, in people's own words	$0.003 per result
forum_search	Google Forums Search	Public forum discussions Google shows for a question	$0.065 per search page
youtube_trending	YouTube Trending Videos	Videos on YouTube's trending charts by country	$0.03 per run + $0.0006 per video
youtube_transcript	YouTube Transcript Extractor	What is said in a video, as text	$0.005 per transcript
youtube_channel_stats	YouTube Channel Statistics	Subscribers, views and activity of channels you might approach	$0.0035 per channel
youtube_comment	YouTube Comments Scraper	What viewers say under a video	$0.001 per comment
creator_report	Creator Intelligence Report	A report on a creator's audience fit and recent posts	$0.05 per creator report
newsletter_post	Substack Posts Scraper	Posts from newsletters on your topic	$0.005 per post
traffic_rank	Website Traffic and Rank	Estimated traffic and rank of a website	$0.08 per domain
tech_stack	Website Technology Detector	What a site is built with	$0.01 per site
bing_rank	Bing Rank Tracker	Where a domain ranks on Bing for your keywords	$0.008 per rank check
crawl_page	Website SEO Crawler	A full-site crawl, page by page	$0.01 per page
local_rank	Local Rank Tracker	Where a business shows on the map around a location	$0.06 per local check
local_pack	Google Local Pack	Who appears in Google's map results for a search	$0.03 per snapshot
agency_listing	Agency Directory Listings	Listings of agencies and companies in a category	$0.004 per listing

Each paid run also pays start fees (usually $0.00005 per GB of run memory; three Actors have none). Apify also bills a small amount of compute on some jobs; it is included in the quote. Prices were read from each Actor's public pricing on 2026-11-08 and can change: the assistant checks again before quoting.

Easy bundles. You do not need to pick jobs yourself. Ask for a pack and the assistant prices it in one go:

Pack	What you get	About
starter	A first look at a website	$0.17
keywords	What people search for on one topic	$1.42
competitors	How rivals look and rank	$0.34
creators	YouTube creators worth approaching	$0.40
local	Where a business shows on the map	$0.15
ai_check	Does AI recommend you (30 questions, 3 assistants, 3 times)	$23.60

Pricing and cost example

• Citebase itself: the twelve tools add no fee of their own. The connector has no start fee and no minimum. Everything that runs costs what Apify charges for compute plus the data jobs below.
• Platform usage: running this Actor uses a small amount of Apify platform compute. It is paid by the account that starts the run, which is yours when you connect with your own sign-in. A 15-page site check takes under a minute.
• Paid data Actors: each is priced per result on your account, as in the table. quote_cost totals them before anything runs.

Example	What it includes	About
Quick audit	1 site audit + 15 results pages	$0.17
Full audit	site audit, keywords, backlinks for you and 3 competitors, link prospects, a small AI check	$8.70
AI recommendation baseline	30 questions x 3 assistants x 3 repeats + 20 Google AI Overviews	$23.60

Request fees dominate small keyword and backlink jobs, so batch them into one run.

Use with an AI agent through Apify MCP

Prompt to try: "Use Citebase to check example.com, tell me the three most important fixes in plain words, and price a Google results check for my top 15 keywords."

• Inputs: every tool field is described, with an example and limits, and job names are a closed list.
• Failures: errors are plain sentences that say what to fix, for example an unknown job lists the valid names.
• Size: results are returned whole. The longest playbook is about 2,300 tokens, so nothing needs paging. For paid Actors use get-dataset-items with limit and offset.
• Run only what was quoted: the assistant should run only the Actor and the spending cap that quote_cost gives you, after your yes.
• No automatic guard on the web: the assistant is told to quote costs and ask you before spending or sending. Say yes only to what you understand.

Limits, reliability and troubleshooting

• Site check: up to 30 pages, public sites only, server HTML only (no JavaScript), about a minute at most.
• Nothing is saved unless you give a project name. Saved projects live in a named key-value store in your own Apify account; up to 100 projects and 60 rounds each.
• Requests are limited to 2 MB.
• Save to one project from one conversation at a time. Two saves to the same project at the same moment from different sessions can overwrite each other.

Symptom	Likely cause	Fix
"Cannot reach" on the site check	The address is private, blocked or wrong	Use a public address such as example.com
"Input should be ..." on quote_cost	The job name is not on the list	Use a name from the table above
"Request is too large"	More than 2 MB was sent	Send fewer rows
401 "api-token-missing"	Connected directly without a token	Use the setup address, or pass your Apify token
ChatGPT will not accept a token	Custom connectors there use OAuth only	Use the setup address and sign in
The assistant says it has no tool to run data jobs	The connector was added without the run tools	Add the connector again using the setup address

Evidence and freshness

Prices were read from public Actor pricing on 2026-11-08. The release checklist covers good, bad, oversized and hostile inputs, parallel calls and all three ways in (Standby, /run and normal runs). It is a check, not a guarantee of future reliability. The version history is in the changelog.

Frequently asked questions

What does it cost? Citebase adds no fee. You pay Apify's running cost and the data jobs you choose to run, each priced before it starts (a first look is about $0.17). Apify also bills a small amount of compute on some jobs; it is included in the quote.

Does it store my data? Only if you ask it to. With a project name, save_results keeps your rounds and site checks in a store named citebase-projects in the Apify account that runs the Actor (yours, when you connect with your own sign-in). The author cannot see it. Without a project name every call stands alone and nothing is kept. Reports are shared by link only: delete the record to stop sharing.

Can it send emails for me? No. It checks a draft and returns a fingerprint of the exact text. You or your mail connector send it.

Which assistants work? Any assistant that supports MCP connectors. ChatGPT needs the OAuth route through mcp.apify.com.

How accurate is the AI check? It uses search-enabled model endpoints, which can differ from what a signed-in person sees. Use the trend over time, not a single number, and run each question at least three times.

Why not just ask the assistant directly? An assistant alone has no live data and no fixed method. Citebase adds the guidelines, a repeatable measurement and the price controls.

Related Actors and workflow steps

• Site problems: Complete SEO Audit for a second opinion on the same site.
• Search results and questions people ask: Google SERP Scraper.
• Do AI assistants mention you: AI Brand Monitor, then AI Search Visibility Tracker for which pages get cited.
• Who links to you and your competitors: Website Backlink Checker.
• Contacts for outreach: Bulk Website Contact Extractor, then Email Address Validator before you send.

Legal, trademarks and responsible use

Use Citebase for honest marketing only. It does not support fake reviews, fabricated press, hidden prompt injection or other manipulation. Its outreach checks are not legal advice: follow email laws such as CAN-SPAM, GDPR and CASL. Do not collect personal data you have no basis to use.

ChatGPT is a trademark of OpenAI. This independent Actor is not affiliated with, associated with, or endorsed by OpenAI.

Claude is a trademark of Anthropic. This independent Actor is not affiliated with, associated with, or endorsed by Anthropic.

Google, Gemini and YouTube are trademarks of Google LLC. This independent Actor is not affiliated with, associated with, or endorsed by Google.

Perplexity is a trademark of Perplexity AI. This independent Actor is not affiliated with, associated with, or endorsed by Perplexity AI.

Reddit is a trademark of Reddit, Inc. This independent Actor is not affiliated with, associated with, or endorsed by Reddit.

Trustpilot is a trademark of Trustpilot A/S. This independent Actor is not affiliated with, associated with, or endorsed by Trustpilot.

G2, Capterra and TrustRadius are trademarks of their respective owners. This independent Actor is not affiliated with, associated with, or endorsed by them.