← All guides marginal.sh

An agent-callable email API for subject line testing

Updated June 2026 · Hosted email MCP server

An agent-callable email API lets your AI assistant run real email experiments without you stitching together SDKs, webhooks, and analytics dashboards. marginal exposes exactly that through a hosted email MCP server — your agent calls a tool, marginal sends and measures.

What 'agent-callable' actually means

A traditional email API expects a human developer to read docs, write request code, and handle auth manually. An agent-callable email API is shaped for autonomous use: the operations are discrete, typed tools that an AI agent can discover and invoke on its own.

marginal delivers this over the Model Context Protocol. Once connected, an agent sees a small, purpose-built toolset rather than a sprawling REST surface, so it can reason about email work step by step.

Why agents make subject line A/B tests easier

Subject line testing is repetitive and metrics-heavy — a good fit for AI agent email marketing. The agent can brainstorm variants, launch the split, wait for signal, and read the lift without a human babysitting each step.

Because marginal handles the send infrastructure and tracking, the agent never touches deliverability plumbing. It reasons in terms of experiments and outcomes, which keeps prompts and tool calls simple.

Connecting the API to your agent

marginal is hosted at https://marginal.sh/mcp, so there's nothing to deploy or self-host. You authenticate with a Bearer API key (marg_live_...) and point a supported MCP client at the endpoint.

Clients include Cursor, OpenAI Codex, Claude Desktop, Claude Code, Windsurf, Cline, Continue, and Zed. Add the server config once and the email tools become available to whatever agent runs inside that client.

Get started with marginal

marginal is a hosted email marketing MCP server at marginal.sh. Sign up free, create an API key, and connect https://marginal.sh/mcp from Cursor, Codex, or Claude Desktop.

Documentation · Sign up · FAQ