Lusha MCP and Linkeddit in Claude: find the contacts, then know what to say

The Lusha connector is the contact side. Lusha's listing describes 300M+ verified contacts with emails and phone numbers plus buyer signals. The Agent Index listing shows 24 tools, including contacts_search, companies_search, decision_makers_search, lookalike_companies and signals_companies_search, while Lusha's own docs say 50 tools and that the set changes as it adds more, so check the tools menu in Claude for the current list.

Linkeddit is the competitor and AI visibility side. Compete tracks a competitor's blog, changelog and monitored pages with a priority on each move, and stores its public reviews from G2, Capterra, TrustRadius and Trustpilot. Answer Radar records which brands the five answer engines name when they answer your buying questions. All of it comes with one plan at $60 a month. Linkeddit holds no contact data and writes no messages.

Lusha answers "who". Linkeddit answers "why them, why now". Claude reads both and drafts the outreach plan, and you decide what is sent.

It adds a reason that is true and dated. A B2B SaaS founder described the wall every cold message hits: most SMBs "already use one of the big incumbents (or a cheap alternative), so cold outreach hits the “we already have something for that” wall immediately." A contact list does not solve that. A specific competitor event does.

On Linkeddit's test project, plausible.io, a read-only run returned two high priority competitor signals for the week of 2026-10-26: Matomo published critical PHP security guidance, and Fathom Analytics launched an AI assistant called Luna. Each came with a source link and a line on why it matters. Claude can give that to Lusha as the profile to search for, for example the people responsible for web analytics at companies that run a self-hosted tool, and open with the real event.

The same run stored public reviews. For Matomo, the monitor read 7 Trustpilot reviews and 20 on G2. A one-star Trustpilot review calls its support slow, and another says its tag manager does not work. A review is one customer's view on one date, so Claude should quote it with its site and date and never as a trend.

Not an alert feed. Claude works when you start it, so a competitor move reaches your prospect list when you run the prompt. Linkeddit's side runs on its own clock: on the test project all five competitor monitors are set to weekly, each with a next run of 2026-11-04, so the data is ready when you ask.

A workable weekly routine is to have Claude read the high priority competitor signals with Linkeddit, then ask Lusha for the buyer signals and decision makers that fit the competitor's customer profile. On the test project the week produced two high priority signals, so a short list is a normal result and the output should say so. If you need a real-time alert, use a tool built to push one.

Start with who actually competes with you in search, then use those names to shape the Lusha search. Linkeddit's competitor footprint on the test project covers 59 checked keywords and 358 competing domains. Matomo ranked in the Google top 10 on 11 of those keywords, and ChatGPT cited matomo.org 21 times in the latest finished answer run.

Claude can take the competitor names from that data, ask Lusha for similar companies with lookalike_companies or for people in a role with decision_makers_search, and read recent signals with signals_companies_search. One boundary is worth stating: a competitor's domain is a vendor, not a prospect, and the listing does not show a tool that lists a competitor's customers. The prompt should tell Claude to say so instead of guessing which companies use a rival.

Sometimes. Answer Radar records which domains the engines cite when they answer your buying questions, and answer_gaps_get lists the questions where an engine recommended somebody else. On the test project the top open gap was "Umami vs Matomo for self-hosted analytics", with an intent score of 0.6 and an evidence score of 22.

Lusha covers people at companies, so it helps when a cited source is a company with a team you can contact, such as a vendor or a publisher with a business entity. It is not a directory of every editor or blogger. Ask Claude to read the cited domains with Linkeddit, look each company up in Lusha, and say plainly which it could not find. Reach out about a useful page, never to buy a citation.

Check credits, consent and claims. Lusha's setup page says the MCP server draws from the same credit pool as its REST API, and its listing marks the connector as read and write with one transactional tool, so tell Claude to read account_usage first and to enrich nothing without your yes. The recommendations tools work only over an OAuth connection, which is how Claude connects.

Contact data is personal data. Lusha's listing says it is backed by GDPR, CCPA, SOC 2 Type II and ISO 27701 compliance, which describes Lusha, not your use of the data, so follow your own policy and the rules of the countries you contact. Verify each competitor claim against its source link before it appears in a message.