Skip to content
Pathbound DOCS

MCP Server — Overview

The Pathbound MCP server exposes the REST API as a set of tools for Model Context Protocol clients. The most common use is connecting Pathbound to Claude.ai so you can ask questions like “what did this contact do this week?” or “show me the visitors who hit /pricing yesterday” — and have the model read your unified customer profiles, events, and activity history. With the tools a workspace admin switches on, it can also act: tag records, save segments, enrich contacts, and draft or send email through the tools you have connected.

https://mcp.pathbound.ai/mcp
Use…When
REST APIYou’re writing your own backend, integration, or scheduled job.
MCP serverYou want Claude.ai (or another MCP client) to read Pathbound data, and act on it, on your behalf.

The two surfaces hit the same backend and respect the same tenant boundaries. Interactive MCP clients (Claude.ai) authenticate via Pathbound’s native OAuth login; headless MCP clients can authenticate with a REST API key instead.

More than 100 tools across these domains:

  • Contacts and companies — list, get, create (one or many), update, timeline, conversations, meetings, email threads, notes.
  • Deals — list, get, create, update, delete, notes.
  • Events — list, get, types, stats.
  • Tags and segments — tag records one at a time or by segment, save segments, and run an action on everyone in one.
  • Batch, aggregate and schema — multi-entity lookups, bulk meeting notes, server-side group-by, and the list of fields you can filter on.
  • Knowledge — your workspace’s own pages about its business, which an agent reads before it writes for you.
  • Streaming — automations that tag records, change a Resend segment, or email everyone who submits a form.
  • Actions in your connected tools — Gmail, Resend, Apollo sequences, Instantly, and enrichment and prospecting through Apollo or Hunter.

The tools that read and write your own records are always available. Every other group is off until a workspace admin switches it on under MCP Tools in the dashboard, and the tools that act in a connected tool also need that integration connected.

See the full Tools Reference.

  1. In Claude.ai, open Customize → Connectors and select + → Add custom connector.
  2. Enter the URL https://mcp.pathbound.ai/mcp and select Add.
  3. Select Connect. Claude opens the OAuth flow — log in with your Pathbound account.
  4. Approve the connection. You’re now ready to ask Claude questions about your Pathbound data.

On a Team or Enterprise plan, an owner adds the connector first under Organization settings → Connectors; each member then connects their own Pathbound account from Customize → Connectors.

The OAuth grant is per-user and respects your Pathbound permissions: anything you can read or write in the Pathbound dashboard is what the MCP tools will let Claude do.

A connection reaches one workspace: the workspace of the account that approved it. To work in several workspaces from the same Claude, add one connector per workspace, approve each one with that workspace’s account, and name the workspace at the end of the connector’s URL:

https://mcp.pathbound.ai/mcp?ws=acmecorp

ws is the workspace’s name in lowercase letters and digits (Acme Corp is acmecorp), or the workspace’s id, which stays the same if the workspace is renamed. Accents, capitals and punctuation make no difference: Müller GmbH is mullergmbh, and müller-gmbh finds it too. The get_usage tool shows both under workspace.

  • The server tells the model which workspace each connection is, so a conversation with several Pathbound connectors can use the right one.
  • If the URL names one workspace and the account that approved the connector belongs to another, the server refuses the connection and says which workspace it is signed in to. That happens when the browser was signed in to Pathbound with another account at the time you approved. On the consent screen, choose use another account before you approve.
  • A URL without ws connects as before, to the workspace of the account that approved it.

Server-side agents — hosted MCP tool runners (OpenAI, Anthropic API), schedulers, CI pipelines — can’t open a browser to complete the OAuth flow. They connect with a REST API key sent as a static bearer header:

Authorization: Bearer sk_your_api_key

Create the key under Settings → API keys at app.pathbound.ai/settings/api-keys; its scopes (e.g. contacts:read) apply to MCP tool calls exactly as they do on the REST API. See Authentication for details.

  • The MCP server is stateless and rate-limited per user. Heavy bursts (more than ~30 tool calls per minute) may be throttled.
  • Tool responses are bounded — get_contact caps at 20 events. Use aggregate_data and batch_get_* instead of paginating.