LinkedIn MCP (by TalkToHumans)

What LinkedIn MCP (by TalkToHumans) does

LinkedIn MCP (by TalkToHumans) gives compatible AI clients access to a TalkToHumans workspace connected to LinkedIn. It supports work across LinkedIn conversations, the user's network, Sales Navigator, and Recruiter threads. Agents can identify likely leads, find conversations that need replies, research contacts, and prepare personalized outreach.

The service also supports contact enrichment using work email and mobile phone data gathered through more than 20 providers. Outreach analysis can cover connection acceptance, reply rates, templates, and progression toward booked meetings. Contacts can be scored against an ideal customer profile, tagged, imported from external lists, or synchronized with a CRM through a CRM MCP or the native HubSpot integration.

How it works

The MCP endpoint is hosted at https://api.talktohumans.app/mcp and uses Streamable HTTP with OAuth. The agent reads a synchronized copy of LinkedIn data rather than issuing a fresh LinkedIn request for every question. LinkedIn actions are carried out in the browser already used by the account, through the connected TalkToHumans extension.

Message creation is review-oriented. The agent can draft a reply or stage a follow-up sequence, but the user must review and send it. Follow-up sequences can be configured to stop when a lead replies. Work can be reviewed in the agent's chat, the TalkToHumans app, or embedded panels in LinkedIn and Sales Navigator.

Setup and configuration

Use of the MCP requires a TalkToHumans account on the Pro plan and the TalkToHumans browser extension connected to LinkedIn. Add the remote endpoint to an MCP client, then complete the TalkToHumans sign-in flow when prompted.

Examples include:

• Claude: add a custom connector named TalkToHumans with the endpoint URL.
• Claude Code: run claude mcp add --transport http --scope user talktohumans https://api.talktohumans.app/mcp.
• Codex: run codex mcp add talktohumans --url https://api.talktohumans.app/mcp, followed by codex mcp login talktohumans.
• Cursor: add the endpoint under mcpServers in ~/.cursor/mcp.json.
• VS Code: use MCP: Add Server, select HTTP, and add the endpoint globally.

The service can also be used from other clients that support remote MCP over Streamable HTTP with OAuth.

Tools and capabilities

• Search a LinkedIn network for contacts matching an ICP.
• Find conversations across LinkedIn, Sales Navigator, and Recruiter where a reply may be due.
• Draft personalized messages and stage follow-up sequences.
• Enrich contacts with work email and mobile phone information.
• Analyze connection acceptance, reply, template, and meeting outcomes.
• Import lead lists, score contacts, and apply contact tags.
• Log LinkedIn conversations in a CRM or use the native HubSpot integration.
• Create working views in TalkToHumans for lead follow-up.

Limitations and notes

LinkedIn does not provide an official MCP server, and its public API does not expose private messages, so this service depends on TalkToHumans and its browser extension. It is not risk-free. Account-level daily limits and gaps between actions are enforced, and the agent cannot raise them. Synchronization and background activity are paced as well.

The MCP drafts messages rather than sending them. Users remain responsible for reviewing and sending outreach. ChatGPT custom MCP servers require a Business, Enterprise, or Edu plan. Team and Enterprise Claude organizations may require an owner to add the connector first.