Version: v3

LeadConnector MCP Server

We’re excited to announce that the LeadConnector MCP (Model Context Protocol) server is live and ready for use! It opens up a world where advanced AI assistants can talk directly to your CRM data and tools. Think of it as a bridge: you can now query, automate, and orchestrate everything in your LeadConnector account with AI.

The best way to connect your agents to LeadConnector is the per-client endpoint https://services.leadconnectorhq.com/mcp/{client}/v2 — live today for Claude at /mcp/anthropic/v2, with more clients on the way.

What is MCP?

MCP is an open protocol that standardizes how applications provide context to large language models. The LeadConnector MCP server gives AI assistants a standardized way to work with your LeadConnector data and operations — without you or the model needing to know the internal details of how the underlying APIs work.

Authentication (both endpoints)

Every LeadConnector MCP endpoint supports both authentication methods — pick whichever your client prefers. In both flows, you choose exactly which scopes to grant the integration:

OAuth (recommended) — one-click sign-in through the LeadConnector consent flow (nothing to store or rotate). You approve the scopes on the consent screen, and OAuth makes the widest set of scopes available.
Private Integration Token (PIT) — create a token under Settings → Private Integrations, selecting the scopes you want it to have, and pass it in the Authorization header. A PIT offers a more limited set of scopes than OAuth.

Because OAuth exposes a broader scope set than a PIT, it unlocks more of the operation catalog — another reason it’s the recommended choice. Either way, the integration can only do what the scopes you grant allow.

Every request targets a single sub-account (location). The original /mcp/ endpoint is one sub-account per connection. The per-client /mcp/{client}/v2 endpoint additionally lets an agency connect once and work across many sub-accounts — choosing the sub-account per request (see Working across sub-accounts).

Which endpoint to use

	`/mcp/{client}/v2` — recommended	`/mcp/` — original
Endpoint	`https://services.leadconnectorhq.com/mcp/anthropic/v2` (live) — more clients planned	`https://services.leadconnectorhq.com/mcp/`
Auth	OAuth or Private Integration Token	OAuth or Private Integration Token
Clients	Per-client; Claude live today, others planned	Any HTTP-based MCP client (Cursor, Windsurf, n8n, custom agents, and more)
Coverage	Widest — the full operation catalog: hundreds of operations across 40 domains	Limited — a focused set of core tools / narrower scope
Level	Single sub-account, or agency-wide across many sub-accounts (one connection)	Single sub-account per connection

Recommendation: use /mcp/{client}/v2 for the widest coverage. It’s live today for Claude at /mcp/anthropic/v2. The original /mcp/ endpoint works with any MCP client (via OAuth or PIT) but exposes a more limited toolset.

Recommended: the per-client endpoint (`/mcp/{client}/v2`)

The recommended way to connect your agents. Each client gets a dedicated endpoint following the pattern https://services.leadconnectorhq.com/mcp/{client}/v2, backed by the full LeadConnector operation catalog.

Live today: https://services.leadconnectorhq.com/mcp/anthropic/v2 (Claude). Other clients are on the roadmap.

▶️ Walkthrough video: Watch on Loom

Widest coverage — the full operation catalog (hundreds of operations across 40 domains), well beyond the original /mcp/ toolset. The exact operations an assistant sees are filtered by your grant.
OAuth or PIT — connect with one-click OAuth (recommended) or a Private Integration Token in the Authorization header.
One sub-account, or your whole agency — connect to a single sub-account (the default), or, as an agency, connect once and work across many sub-accounts from the same connection. Every request still runs against one sub-account you choose — see Working across sub-accounts.
A small, stable toolset — instead of hundreds of individual tools competing for the model’s attention, the server exposes a compact set of unified tools (below). The assistant uses them to discover and run everything else.

Single sub-account: with OAuth, your client opens a browser to the LeadConnector sign-in page — sign in, choose the sub-account (location) to expose, and approve. The connection then operates on that location for its lifetime.

Agency, many sub-accounts: installing as an agency, you choose which sub-accounts to include. The one connection can then work across all of them — the assistant uses the sub-account you name, or asks which one when it’s ambiguous, and you can switch between them in the same chat. It only ever touches the sub-accounts you selected.

A small, unified toolset

Rather than listing every operation as its own tool, the server presents a handful of unified tools. There are only a few to learn — the assistant handles the rest.

Tool	What it does
`search`	Find customer or business records by name, email, phone, tag, or similar criteria
`fetch`	Retrieve the full details of one or more records returned by `search`
`search_operations`	Discover the available operations by intent — list, create, update, delete, and more
`describe_operation`	Inspect an operation’s inputs before running it
`execute_operation`	Run one operation, subject to your scopes and built-in safety checks
`list_locations`	List the sub-accounts this connection can use, so you can pick one (agency connections)

Behind these sits the full LeadConnector operation catalog — hundreds of operations across 40 domains. Use search_operations at any time to see what’s available to your grant.

A single prompt in action

Once connected, you don’t need to know the LeadConnector API — you just ask. For example:

“Find the contact with email [email protected], add the tag ‘vip-2026’, and create a new opportunity in the ‘Sales Pipeline’ for them worth $5,000.”

Behind the scenes the assistant:

search → finds Jane’s contact
search_operations → discovers the “add tags” and “create opportunity” operations
describe_operation → inspects the inputs each operation expects
execute_operation → adds the tag and creates the opportunity

You get a single, natural-language confirmation back.

Working across sub-accounts (agencies)

▶️ Walkthrough video: Multiple sub-account support in Claude.ai — watch on Loom

Individual users connect to one sub-account and never think about locations again. Agencies can instead connect once and work across many sub-accounts from the same connection — no separate connection per sub-account.

How it works:

Install for your agency. When you connect, choose which sub-accounts to include (one, several, or all). The connection can work with exactly those — and no others.
Name the sub-account, or let the assistant ask. In a request, refer to the sub-account by name (e.g. “in Downtown Clinic, …”). If you don’t say which one and it’s ambiguous, the assistant calls list_locations and asks you to choose.
Switch freely in the same chat. Ask about one sub-account, then another, without reconnecting — “now do the same for the Uptown location.”

Example:

“List the contacts tagged ‘vip-2026’ in Downtown Clinic, then compare the count with Uptown Med Spa.”

The assistant uses list_locations to resolve the names, runs the query against each sub-account, and reports both — each request authorized against your installation.

One sub-account at a time, always authorized. Each request runs against a single sub-account you chose at install. A request for a sub-account you didn’t include (or that isn’t part of the installation) is refused — the connection can never reach beyond the sub-accounts you selected.

Availability: individual single-sub-account connections work today. Agency-wide connections across multiple sub-accounts are rolling out — an agency enables them by installing with the multi-sub-account (bulk) option. If your install offers only a single sub-account, that option isn’t enabled for your app yet.

What you can do

Coverage spans your whole LeadConnector account, including:

Contacts — get, list, search, create, update, delete, upsert, duplicate lookup, tags, notes, tasks, followers, appointments, and business assignment
Conversations & Messages — create, get, update, delete, and search conversations; read and send messages; message status
Opportunities & Pipelines — pipelines, lost reasons, search, create, update, delete, upsert, followers, and status changes
Calendars & Appointments — calendars, groups, appointments, events, notes, notifications, free/blocked slots, resources, schedules, services, and service bookings
Payments — coupons, integrations, orders, order fulfillment, subscriptions, and transactions
Products & Store — products, prices, collections, inventory, reviews, shipping carriers/zones/rates, and store settings
Invoices & Estimates — invoices, estimates, templates, schedules, and send/void actions
Social Planner — accounts, posts, categories, tags, calendar views, and statistics
Blogs — blog sites, authors, categories, posts, and slug checks
Emails — templates, template folders, campaigns, scheduling, and campaign stats
Forms & Surveys — forms, surveys, and their submissions

…and more. Ask search_operations for the exact operations available to your grant.

Connect Claude (`/mcp/anthropic/v2`)

Claude connects to https://services.leadconnectorhq.com/mcp/anthropic/v2. One-click OAuth is recommended; a Private Integration Token also works.

Claude.ai

Open Claude.ai → Settings → Connectors → Add custom connector.
Set the server URL to https://services.leadconnectorhq.com/mcp/anthropic/v2.
Click Connect and complete the LeadConnector sign-in (sign in → pick sub-account(s) → approve).
Start a new chat — the LeadConnector tools are now available.

Tip: Ask “Find the last 5 contacts I added in LeadConnector” to confirm the connection works.

Claude Code (CLI)

claude mcp add --transport http leadconnector https://services.leadconnectorhq.com/mcp/anthropic/v2

Or add it to .mcp.json at your project root:

{
  "mcpServers": {
    "leadconnector": {
      "type": "http",
      "url": "https://services.leadconnectorhq.com/mcp/anthropic/v2"
    }
  }
}

On first use, Claude Code opens a browser for LeadConnector authorization. Verify with claude mcp list.

Claude Cowork

In Claude Cowork, open the connectors / integrations settings and choose Add custom connector.
Set the server URL to https://services.leadconnectorhq.com/mcp/anthropic/v2.
Complete the LeadConnector sign-in (sign in → pick sub-account(s) → approve).
Your Cowork agents can now use the LeadConnector tools.

Prefer a token? Instead of OAuth, you can pass a Private Integration Token in the Authorization header (Bearer pit-your-token) — see the PIT setup note above.

Permissions & security

You choose the sub-accounts. At connect time you select which sub-account(s) the connection can operate on — a single one, or (for agencies) several. It never reaches a sub-account you didn’t include, and every request is authorized against your installation before it runs.
Scoped to your grant. The assistant can only perform operations your OAuth scopes (or PIT scopes) allow. With OAuth you can review or revoke access at any time from your LeadConnector account.
Safety checks. Sensitive and irreversible operations are gated with additional confirmation and safety checks before they run.

Also available: the original endpoint (`/mcp/`)

Endpoint: https://services.leadconnectorhq.com/mcp/

▶️ Walkthrough video: Watch on Loom

The original LeadConnector MCP endpoint. It works with any HTTP-based MCP client (Cursor, Windsurf, n8n, or your own agent), supports both OAuth and Private Integration Token auth, and is a sub-account (location) level feature. It exposes a more limited, focused set of core tools (contacts, conversations, opportunities, calendars, payments, social planner, blogs, emails) with a narrower scope than the /mcp/{client}/v2 catalog.

For the widest coverage, use the recommended per-client endpoint above.

Connect with OAuth: point your MCP client at https://services.leadconnectorhq.com/mcp/ and complete the sign-in when prompted.

Connect with a Private Integration Token: create a token under Settings → Private Integrations (selecting the scopes your AI needs), then add the endpoint and headers to your client. The locationId header is optional — it can also be provided in prompts.

{
  "mcpServers": {
    "leadconnector-mcp": {
      "url": "https://services.leadconnectorhq.com/mcp/",
      "headers": {
        "Authorization": "Bearer pit-your-token",
        "locationId": "your-location-id"
      }
    }
  }
}

Roadmap

The recommended per-client OAuth experience is delivered per client, each with its own endpoint following the pattern:

https://services.leadconnectorhq.com/mcp/{client}/v2

Available now — /mcp/anthropic/v2 (Claude: Claude.ai, Claude Code, Claude Cowork).
Planned & coming soon — a dedicated /mcp/{client}/v2 endpoint for:
- OpenAI (ChatGPT & Codex)
- Cursor
- Windsurf
- VS Code
- …and more.

Until a client’s dedicated endpoint ships, it can connect today via the original /mcp/ endpoint (OAuth or PIT). Check back here as each becomes available.

Try it out & feedback

Please try out the MCP server and let us know what you think — we value your feedback!
Connect Claude, or build your own agent on either endpoint.
For questions or support, feel free to reach out.

What is MCP?​

Authentication (both endpoints)​

Which endpoint to use​

Recommended: the per-client endpoint (/mcp/{client}/v2)

A small, unified toolset​

A single prompt in action​

Working across sub-accounts (agencies)​

What you can do​

Connect Claude (/mcp/anthropic/v2)​

Claude.ai​

Claude Code (CLI)​

Claude Cowork​

Permissions & security​

Also available: the original endpoint (/mcp/)

Roadmap​

Try it out & feedback​