CitedBy MCP¶
The audit that asks the only question that matters now: when a person asks an assistant about your market, is your page readable and does it get cited.
Server type: Remote (Streamable HTTP)
Endpoint: https://citedby.eu/mcp
Transport: JSON-RPC 2.0 over HTTP
Auth: none - every tool is free and read-only
Tools: 4
Built by: CitedBy (citedby.eu)
Registry: not listed in the official MCP registry
Why This Matters for Operators¶
Search visibility was a ranking problem and is now a readability problem. An assistant answering a question reads a handful of pages, and it can only cite the ones it was allowed to fetch, could parse, and could attribute. A site that blocks GPTBot or ships a page whose structured data Google rejects is invisible in a way that never shows up as a traffic drop you can explain.
Most AEO tooling sells a score. This one returns a finding, the reason it matters, and the exact fix, and it refuses to score what it could not read.
What It Actually Does¶
The endpoint answered without a key in this sweep. initialize returned HTTP 200 as citedby 1.0.0 on protocol 2025-06-18, and tools/list returned four tools in 2.9 KB of schema.
| Tool | What it returns |
|---|---|
check_page |
Fetches one page as CitedBot with robots.txt respected: score, findings, and for each finding the severity, the check name, why it matters and the fix. Deterministic checks only. |
check_ai_crawler_access |
For GPTBot, OAI-SearchBot, ChatGPT-User, PerplexityBot, ClaudeBot, Google-Extended, Bingbot and Googlebot: allowed by robots.txt, and answered by the server, or not. |
check_llms_txt |
Whether /llms.txt exists, whether its links resolve, whether /llms-full.txt exists. Drafts one from the site's own pages when there is none. |
backlinks |
How many domains link to a domain and the strongest referrers, from the Common Crawl domain web graph. |
All four were executed against corpusiq.io in this session, which makes the output concrete.
check_ai_crawler_access returned 8 of 8 crawlers can read the site, dated 2026-10-26, robots.txt okay, each line naming what that bot does: GPTBot for model training, OAI-SearchBot for citing you in answers, ChatGPT-User for opening a page when a person asks about it, PerplexityBot for citing you, ClaudeBot for training, Google-Extended for Gemini grounding, Bingbot and Googlebot for citing you.
check_llms_txt returned present: true, status: 200, 11,256 bytes, 10 links checked, 0 broken, and /llms-full.txt answering 200.
check_page on https://www.corpusiq.io/ returned score 90/100 with 2 findings, both real:
schema_rejected(warning):SoftwareApplicationis missingoffers.price, one ofaggregateRatingorreview. The response carries the consequence in plain words: the markup parses, but a type Google builds results from is failing its rich-result check, and schema is not a ranking lever, so a rejected block is an error in Search Console rather than a penalty. The suggested fix names the alternative: a subscription service is aService, not aProduct.render_blocking(notice): 0 scripts and 7 stylesheets block rendering.
backlinks on corpusiq.io returned in_data: false and said so plainly rather than reporting zero: the dataset covers domains with 100 or more referring domains under .eu, EU and UK country domains, 264,847 domains, plus the most-linked sites, from the Common Crawl graph cc-main-2026-aug-sep-oct read on 2026-10-23. A domain absent from that data says nothing about how many links it has, and the tool states that.
What It Does Not Do¶
- No whole-site crawl on the free tier. One page per
check_pagecall; every page with answer analysis needs a plan at citedby.eu/plans. backlinksis Common Crawl open data, so a US small business is usually outside the coverage window. It is a sanity check, not an Ahrefs replacement.- Nothing is connected and nothing is written. It reads public pages.
- The page score is the sum of deterministic checks. It does not measure whether the writing answers a question, only whether the page is structurally citable.
check_pagerespects robots.txt, so a site that blocks the crawler returns that fact instead of a score.
FAQ¶
Is a rejected schema block actually hurting us?¶
Not as a ranking signal. It hurts because a rich result you could have shown is being suppressed and reported as an error, which hides real breakage in a report you have to read anyway. Fix the missing property or change the type.
Why does crawler access matter more than the score?¶
A blocked crawler means no citation at all, which the score cannot express. Eight of eight open is the only acceptable answer for a business site that wants to be quoted in AI answers.
What is llms.txt good for here?¶
It is the page an assistant reads to know what your site contains and where the canonical pages live. A tool that checks the links and the llms-full.txt companion turns a hand-maintained file into something verifiable.
Can this run on a schedule?¶
Each tool is a single request with a stable shape, so yes: crawler access and the page audit are cheap enough to check weekly and diff, which is how a robots.txt edit by someone else gets caught before the citations disappear.
The Operator Test¶
Point check_ai_crawler_access at your own domain first. If any line reads "cannot read the site", every AEO effort is decoration until that is fixed. Then run check_page on the page that should win your money question and read the two findings.
Pairs with creavores-aeo-mcp and the AI visibility entries already in this catalog, and with the site-health entries, because a page that cannot be fetched cannot be cited.
Integration with CorpusIQ¶
CorpusIQ is the connector layer that lets AI assistants read the systems a business already runs on - accounting, payments, CRM, email, files and analytics - and answer with the source named. It is read-only by design: it retrieves data and never writes to, or acts in, a connected system, which is what makes it safe to point at live accounts.
This page is part of the CorpusIQ directory, the reference we maintain for the MCP server and agent-skill ecosystem. Run alongside CorpusIQ, a third-party server like this one can be combined with a company's own business data in a single answer.