MCP server documentation

Connect your agent to every Vayne tool

The Vayne MCP server lets AI agents — Claude, Cursor, and any MCP-compatible client — use every Vayne API feature as tools: Sales Navigator scraping, profile and company scraping, job and post scraping, reverse email lookup, and email & phone enrichment. This page covers connection details, per-client setup, the tool catalogue, credits, and webhooks.

Discover the Vayne MCP

Transport: Streamable HTTP, hosted and stateless
Authentication: OAuth sign-in, or a Bearer API token
Plans: API access from Starter, same credits
Tools: 30+ tools, discovered automatically

Setup

Connection details

The Vayne MCP server is a hosted, stateless HTTP service — nothing to install or run yourself.

Transport: Streamable HTTP
Auth header (token clients only): Authorization: Bearer <YOUR_TOKEN>

Server URL

https://mcp.vayne.io/mcp

Two ways to authenticate. OAuth is the default for Claude: Claude Desktop, Claude.ai, Claude Code and Cursor only need the server URL — they redirect you to Vayne to sign in and authorize, and store no token on your machine. To end an OAuth connection, remove the connector in the client; the access tokens it holds are short-lived and expire on their own. Everything else sends a Vayne API token in the Authorization header.

Using the MCP requires a plan with API access (Starter and above), OAuth connections included: on a lower plan the connector still authorizes, but every tool call comes back with 'Plan does not include API access'. Get or rotate your API token in the Vayne app under API Settings; logged-in users will also find ready-made client configurations in the MCP Setup modal on any feature page.

Give this documentation to your AI agent

The full agent guide — every tool, the credit model, webhooks, and export column glossary — is published as a single markdown page. Copy it and paste it into any AI assistant, or hand the raw URL to your agent.

View raw markdown

Pro Tip: MCP clients discover the tools automatically when they connect — the guide is extra context that helps your agent pick the right tool and estimate costs before spending credits.

Client setup

Pick your client below. Claude Desktop, Claude.ai, Claude Code and Cursor can sign in to Vayne with OAuth — nothing to copy but the server URL. The remaining snippets expect your API token in place of <YOUR_TOKEN>.

Claude Desktop (recommended)

Claude Desktop talks to the Vayne server directly. There is no config file to edit, no API token to copy and nothing to install — just the server URL and your Vayne login.

In Claude Desktop, open Settings → Connectors.
Click Add custom connector at the bottom of the list.
Name it Vayne and paste the server URL below.
Click Add, then Connect — Claude opens Vayne in your browser.
Sign in to Vayne if you are not already, then click Authorize.
Back in Claude, open a new chat and ask “how many Vayne credits do I have?” to confirm the tools are live.

Server URL to paste

https://mcp.vayne.io/mcp

Custom connectors require a paid Claude plan (Pro, Max, Team or Enterprise). Vayne never sees your Claude account, and no API token is stored on your machine. To disconnect later, remove the connector in Claude — it then has nothing left to call Vayne with.

Advanced: connect with an API token instead

Only useful if custom connectors are unavailable on your Claude plan. This route runs a local bridge and therefore needs Node.js installed. Add this block to claude_desktop_config.json (Settings → Developer → Edit Config), replace <YOUR_TOKEN>, save, then restart Claude Desktop:

{
  "mcpServers": {
    "vayne": {
      "command": "npx",
      "args": ["-y", "mcp-remote", "https://mcp.vayne.io/mcp", "--header", "Authorization: Bearer <YOUR_TOKEN>"]
    }
  }
}

Claude.ai (web & mobile)

Identical to Claude Desktop: Settings → Connectors → Add custom connector, paste the server URL above, then authorize Vayne when Claude redirects you. A connector added on one Claude surface shows up on the others too.

Claude Code (CLI)

Run this in a terminal, then type /mcp inside Claude Code and pick Authenticate to sign in to Vayne in your browser:

claude mcp add --transport http vayne https://mcp.vayne.io/mcp

Prefer an API token? Use this command instead

Replace <YOUR_TOKEN> with your Vayne API token — no browser sign-in, handy for servers and CI:

claude mcp add --transport http vayne https://mcp.vayne.io/mcp --header "Authorization: Bearer <YOUR_TOKEN>"

Cursor

In Cursor, open Settings → MCP and add a new server with this configuration (replace <YOUR_TOKEN>). To use OAuth instead, remove the headers block — Cursor will then prompt you to sign in to Vayne.

{
  "mcpServers": {
    "vayne": {
      "url": "https://mcp.vayne.io/mcp",
      "headers": { "Authorization": "Bearer <YOUR_TOKEN>" }
    }
  }
}

Claude Agent SDK

For building your own agent with the Claude Agent SDK. There is no browser step here, so use a token: replace <YOUR_TOKEN> with the Vayne token of the user the agent acts for.

import { query } from "@anthropic-ai/claude-agent-sdk";

const result = query({
  prompt: "Find CEOs in North America and start a Sales Navigator scrape",
  options: {
    mcpServers: {
      vayne: {
        type: "http",
        url: "https://mcp.vayne.io/mcp",
        headers: { Authorization: "Bearer <YOUR_TOKEN>" },
      },
    },
  },
});

Any other MCP client

Point any MCP-compatible agent (for example Goose, Cline, or OpenCode) at the server URL above. Clients that implement the MCP OAuth flow are redirected to Vayne to sign in; the others send your API token in the Authorization header.

Reference

Tools

30+ tools grouped by feature — the same capabilities as the Vayne API, plus MCP-only helpers like the Sales Navigator search builders.

Sales Navigator Build lead or account searches from plain-English criteria, estimate the result count for free, then create and track scraping orders (up to 20,000 results per search). build_search_url · build_account_search_url · estimate_search · create_order · list_orders · get_order

Profile & company scraping Scrape any LinkedIn profile or company URL — a single URL returns instantly, two or more become an async batch job. scrape_linkedin_url · get_linkedin_scraping

Job scraping & job search Scrape job postings (single or batch) and search LinkedIn jobs by keywords, location, or company, loading extra pages on demand. scrape_linkedin_job · get_linkedin_job_scraping · search_linkedin_jobs · get_linkedin_job_search · load_more_linkedin_job_search

Post scrapers Capture every liker and commenter on one post, or across all recent posts of a profile or company page — with cost estimates first. scrape_post_engagement · scrape_profile_posts · estimate_profile_posts · scrape_company_posts · estimate_company_posts · list_post_scrapers · get_post_scraper

Reverse email lookup Turn B2B email addresses into LinkedIn profiles. Charged per match found, not per email submitted. create_reverse_lookup · list_reverse_lookups · get_reverse_lookup

Email & phone enrichment Find a contact's business email and phone number — one contact per call, identified by LinkedIn URL or name + company domain. Uses enrichment credits. create_contact_finder · list_contact_finders · get_contact_finder

Find people Quick people search by company and job title, without building a Sales Navigator URL. find_people · get_find_people

Account state & helpers Check credit balances and daily limits, verify the LinkedIn connection, and resolve free-text industries, locations, and companies into LinkedIn IDs for precise targeting. get_credits · check_linkedin_status · search_industries · search_locations · search_companies

Credits

MCP actions consume the same credits as the web app and the API — there is no separate MCP pricing. Two pools, both visible via get_credits:

Scraping credits fund Sales Navigator orders, reverse lookups, URL and job scrapings, job searches (including load_more), and post scrapers. Daily limits apply only to Sales Navigator scrapes.
Enrichment credits fund email & phone enrichment (contact finder) only.
Agents are instructed to estimate result counts and confirm credit cost with you before creating an order.

Webhooks

Most create tools accept an optional HTTPS webhook URL that receives the results when the job finishes — handy for Clay, Bitscale, CRMs, Zapier, n8n, or custom backends. Deliveries are protected by trust-on-first-use: the first time a job targets a new destination host, the job runs normally but the delivery is held and you receive a one-click approval email. Once approved, the host is trusted for future deliveries; held deliveries expire after 7 days if never approved.

Good to know

Scraping orders are throttled to 5 per 15 minutes per user, and orders larger than your available credits are refused.
A single Sales Navigator search scrapes at most 20,000 results — larger targets get split into several searches.
create_order can also save the search inside Sales Navigator (savedSearch) — leads only, and LinkedIn caps saved searches at 50 per account.
Sales Navigator scraping uses the LinkedIn session connected to your Vayne account; if it expires, refresh it in the app's LinkedIn connection settings. Other features need no LinkedIn account.
A 401 on any tool means the API token is wrong or was rotated — copy a fresh one from API Settings. On an OAuth connector, remove it and add it again to re-authorize.
MCP clients cache the tool list at connect time; if new tools don't show up, fully disconnect and reconnect the client.

Discover the Vayne MCP REST API documentation