--- title: "How to Connect Claude to LinkedIn for Outreach (2026)" url: https://trysumora.com/connect/claude-linkedin markdown-of: https://trysumora.com/connect/claude-linkedin --- # How to Connect Claude to LinkedIn for Outreach (2026) [Home](/)/[Connect](/connect) Connect # How to connect Claude to LinkedIn Claude can reach LinkedIn through MCP servers and automation bridges. What each approach actually does, where they stop, and what running safe outreach volume additionally requires. By Devansh Ranjan · Published Aug 7, 2026 **Short answer:** Claude connects to LinkedIn through a third-party **MCP server** or an **automation bridge**. LinkedIn publishes no public outreach API, so every option runs on a provider holding your session. Setup takes minutes; **safe pacing is the part that does not come with it**. ## Option 1: a hosted MCP server Several LinkedIn outreach vendors now publish MCP servers. You create an account, connect your LinkedIn seat, copy an API key, and add the server in Claude’s connector settings. From there you can ask Claude in plain language to pull leads, draft messages, tag people, or push prospects into a sequence. Because MCP is a shared protocol, the same server usually works in ChatGPT and Gemini as well. **The assistant is the least important part of this decision.** What matters is the provider behind the server, because that is what holds your account. ## Option 2: an automation bridge Tools like PhantomBuster, Expandi, Lemlist, and Dripify are commonly wired up through Zapier or n8n: Claude writes, the bridge hands off, the outreach tool sends. More moving parts, more control, and the same underlying constraint. ## What neither option gives you Both handle _drafting_ and _triggering_. Neither handles _pacing_ — and pacing is what determines whether the account survives the quarter. A chat window has no memory of how many invitations the seat has already sent this week. Ask for fifty and a naive connector will attempt fifty. Sustained invite volume is the single biggest driver of LinkedIn restrictions, and the ceilings are lower than most people expect: - **~20 connection requests/day** on a warmed free account (35 on Premium) - **~100 per week** is the widely observed soft limit on free seats - **5 per day** for a brand new account, ramping over ~12 days Full tables: [LinkedIn connection request limits](/guides/linkedin-connection-request-limits). ## Option 3: an agent that runs outreach, with MCP on top The third shape separates the two jobs. Outreach runs continuously as a system — discovery, scoring, sequencing, follow-up, all under enforced per-seat ceilings — and the MCP connector is how you _supervise_ it from Claude rather than how you drive each action. This is how [Sumora](/)works. Connect your LinkedIn and Instagram accounts and describe your ideal customer, and the agent runs the funnel under the seat’s [warm-up ramp](/guides/linkedin-account-warmup)and weekly ceiling. Then add Sumora’s MCP server in Claude’s connector settings: ``` https://mcp.trysumora.com/workspace/mcp ``` From Claude you can then review pending approvals, read the unified inbox, query your business brain, and delegate to agents. Your workspace key comes from Settings → API Keys in the dashboard; full setup at [trysumora.com/mcp](/mcp). The same server works in ChatGPT, Gemini, Cursor, and Windsurf. **The important difference:**the ceilings are enforced in the engine, not in the client. Asking Claude for two hundred connection requests does not produce two hundred — the seat’s daily cap still binds. A connector that would simply obey is the riskier design, not the more capable one. ## Frequently asked questions Does Claude have a native LinkedIn integration? No. Claude reaches LinkedIn through a third-party MCP server or an automation bridge you add yourself. Several outreach vendors publish hosted MCP servers that connect in a couple of minutes through Claude's connector settings. Can Claude send LinkedIn connection requests and messages? Through an MCP server from a provider that holds your LinkedIn session, yes. LinkedIn publishes no public API for outreach, so the provider is driving an authenticated session. What varies enormously between providers is whether they enforce safe daily and weekly ceilings. Does the same MCP server work with ChatGPT and Gemini? Usually yes. MCP is a shared protocol, so a hosted server added to Claude's connector settings generally works in ChatGPT and Gemini connector settings too. The assistant you use matters far less than the provider behind the server. Is an MCP server enough for LinkedIn outreach? It is enough to draft and trigger sends. It is not enough to run outreach safely, because the assistant has no memory of how many invites the account already sent this week. Sustained volume is the biggest driver of restrictions, so pacing has to live in the system holding the account, not in the chat window. ---- _Markdown twin auto-generated from the HTML page at build time. Source of truth: the HTML/_next data._