The Moz MCP server in Claude: what Moz Data and Moz Local do, and what Linkeddit adds

Moz's guide says its Model Context Protocol server lets Claude Desktop and coding agents such as Claude Code call the Moz APIs as tools, so you ask in plain language and the client picks the Moz tools. It exposes two endpoints, split by product, and each requires a subscription to the matching product. The guide marks the server as beta, with access limited to account owners for now.

Moz Data (https://api.moz.com/mcp/v1/data) needs a Moz Data API subscription. Its listing shows 26 tools: site metrics, Brand Authority, backlinks and linking domains, keyword metrics, search intent, keyword suggestions, ranking keywords and a quota lookup. Moz Local (https://api.moz.com/mcp/v1/local) needs a Moz Local subscription and lists 32 tools for locations, groups, listing health and local performance. Most teams need only Moz Data.

Authentication is OAuth inside the client, which Moz recommends, or a bearer token that is your Moz API key from the Moz API dashboard. The same key works on both endpoints.

Moz Data covers keyword research, rank checks and link research from one connector. Tools in its listing include DataKeywordMetricsFetch, DataKeywordSuggestionsList and DataKeywordSearchIntentFetch for keywords, DataSiteRankingKeywordList and DataSiteRankingKeywordCount for the terms a domain ranks for, and DataSiteLinkList, DataSiteLinkingDomainList and DataSiteLinkIntersectFetch for links. These read Moz's index. They are not a rank tracker you configure for your own list.

Moz's own MCP dashboard post shows the pattern: one prompt asking Claude to compare bose.com with jbl.com and sennheiser.com ran a competitive snapshot, a follow-up asked for a backlink gap, and a final request turned the files into one HTML dashboard. The post's author calls the link intersect endpoint the best for link gaps because it finds URLs that link to competitors but not to you.

For keyword research in Claude, ask for keyword metrics and suggestions on a seed term, then have Claude group the results by intent. Moz supplies volume and difficulty style data. Linkeddit adds the Google results and the AI Overview recorded for the questions you track, so you can see which keywords already show an answer box that cites someone else.

Linkeddit reports the answer engines and competitor activity, which Moz's tool list does not include. Moz Data's listing has no tool for what ChatGPT, Gemini, Perplexity, Claude or Google AI Overviews recommend. Linkeddit records that for your buying questions, with the cited sources, and tracks competitor moves from blogs, changelogs and monitored pages. It all comes with one plan at $60 a month.

Here is a read-only run on Linkeddit's test project, plausible.io. search_competitors_get returned 358 competing domains from 59 checked keywords. matomo.org ranked in the Google top 10 on 11 of them at an average position of 3.73, beat the project on 8, and was cited 23 times by the answer engines in the latest finished run: 21 from ChatGPT and 2 from Google AI Overviews. umami.is was cited 17 times and posthog.com ranked on 12 keywords.

That list is the input Moz is good at consuming. Hand Claude those domains and ask Moz Data for their metrics, ranking keywords and links. Moz explains why a rival's pages are strong. Linkeddit shows where the rival is actually being named to buyers.

The need is real. "I’m curious how other SEOs are tracking AI visibility without turning it into a spreadsheet of manually checked prompts," asked an SEO consultant on Reddit. A recorded answer history is the alternative to that spreadsheet.

Moz Local is for businesses with physical locations. Its listing describes managing locations and groups, publishing listings to the networks Moz Local syncs to, flagging field level discrepancies and network sync status, and pulling Google Business Profile impressions and leads, Local Grid rank maps and keyword ranking history. Tools include LocalLocationNetworkFieldIssueList, LocalLocationLocalGridHistoryFetch and LocalLocationRankingHistoryFetch.

Some Moz Local tools change live data. The listing names LocalLocationCreate, LocalLocationUpdate, LocalLocationClose and LocalLocationCancel among its writes, so ask Claude to read first and confirm before it calls one. A software company with no storefronts can skip this endpoint. A multi location brand can use it for listing audits, with Linkeddit checking whether the engines name its locations when buyers ask local questions.

Keep each connector to its own job and put the source next to every number. Moz numbers describe links, authority and its keyword index. Linkeddit numbers describe measured Google results, AI answers and competitor activity on your project. Ask Claude to label each figure with the connector it came from and not to blend them into one score.

A workable order is to start with Linkeddit to find the competitors and questions that matter, use Moz Data for the link and keyword depth on those domains, and finish with one report. The competitor analysis guide on this site gives the exact prompt. Claude acts when you start it, so a weekly report means running the same prompt again.