Platform · Connectors · MCP

Native to the Model Context Protocol.

Draft & Goal speaks MCP both ways: a first-party server that exposes the platform to any MCP client, a catalog of native integrations with managed auth, and the option to connect any custom server — all as governed tools.

Book a demo → Read the MCP docs

Draft & Goal MCP

Native · first-party MCP server

Draft & Goal MCP

Our first-party MCP server. Expose Draft & Goal agents, workflows, and knowledge to any MCP client — Claude, Cursor, or your own — and let them run governed marketing work with full traceability.

First-party toolsTrigger workflows, query runs, and read knowledge bases as native MCP tools.

Governed by defaultEvery call inherits the same roles, quotas, and audit trail as the platform.

Any MCP clientWorks with Claude Desktop, Cursor, and any Model Context Protocol host.

Native MCP integrations Integrated auth

A catalog of pre-built connectors with managed authentication — connect in a click, no server URLs or tokens to manage.

8 integrations

Amplitude Product analytics — events, user behavior, and retention insights

Atlassian (Jira & Confluence) Project management and documentation with Jira and Confluence

Content Square Analytics and UX insights powered by Contentsquare

Fathom AI meeting notetaker — transcripts, summaries, and action items

Linear Issue tracking and project management for software teams

Notion Search, read, and create pages in your Notion workspace

Sentry Error tracking — query issues, errors, and project diagnostics

Stripe Payment processing — manage invoices, payments, and customers

The catalog grows regularly. Don't see one you need? Connect it as a custom server below, or see the MCP servers documentation.

Connect any MCP server

Beyond the catalog: point an agent at any MCP-compatible server — Slack, your data warehouse, or your own internal API — with your own authentication.

Add a server in Builder → MCP Servers with its URL and transport, test the connection, and its tools appear in the Tools section of any AI Agent node — where the agent decides when to call them.

NameA friendly label for the server, e.g. “Linear – Issue Tracking”.

Server URLThe MCP endpoint, e.g. https://mcp-server.example.com.

TransportThe protocol. SSE (Server-Sent Events) is supported today.

HeadersOptional auth headers, e.g. Authorization: Bearer your-api-key.

Add the serverIn Builder → MCP Servers, click Add MCP Server and enter the name, URL, transport, and any auth headers.

Test the connectionVerify the endpoint is reachable, the headers are correct, and the server speaks valid MCP — then save.

Use the toolsThe server’s tools appear in the Tools section of any AI Agent node. The agent decides when to call them.

Get started

Connect your stack over MCP.

Point an agent at an MCP server and it inherits your governance, quotas, and audit trail automatically. Bring the workflow — we'll wire the servers.

Book a demo → All connectors