An MCP Server for Email Marketing
If you've been searching for an MCP server for email, marginal is a hosted option built specifically for subject line testing and managed sends — exposed to your AI agent through the Model Context Protocol.
What an email MCP server actually does
An email MCP server gives an AI agent a structured way to run real email marketing tasks instead of just drafting copy. marginal speaks the MCP protocol over a single hosted endpoint, so any compatible client can call its tools directly.
Rather than wiring up SDKs and webhooks by hand, your agent invokes named tools and gets back typed results — variants, test IDs, and metrics it can reason about on the next turn.
- generate_variants — produce subject line candidates for a campaign
- launch_test — start a subject line A/B test with a managed send
- get_results — pull open and click tracking plus within-test lift
- recommend_next — get a data-backed suggestion for the next step
Why marginal over a DIY setup
Building your own email MCP server means handling sends, tracking pixels, statistical comparisons, and auth before you write a single useful agent prompt. marginal is hosted, so there's nothing to deploy or self-host.
You authenticate with a Bearer API key (marg_live_...) against https://marginal.sh/mcp and start running experiments. The free tier covers 100 experiments per month, which is enough to validate the workflow end to end.
- Hosted endpoint at https://marginal.sh/mcp — no infrastructure
- Bearer key auth, no email platform credentials required to start
- Open/click tracking and lift metrics computed for you
- 100 experiments/month on the free tier
Subject line A/B testing through an AI agent
The core loop for AI agent email marketing is tight: ask the agent for subject line variants, launch a subject line A/B test, wait for engagement, then read the lift. Because each step is a tool call, the agent can run the whole cycle without you switching tools.
A typical exchange looks like generate_variants → launch_test → get_results → recommend_next. The agent keeps the test ID in context and uses the within-test lift to decide whether to ship a winner or iterate on new angles.
Works with the MCP clients you already use
marginal connects to the standard MCP ecosystem, so you can drive it from your editor or desktop agent. Add the endpoint and key once and the tools appear alongside your other capabilities.
Supported clients include Cursor, OpenAI Codex, Claude Desktop, Claude Code, Windsurf, Cline, Continue, and Zed. Full setup steps live in the docs at https://marginal.sh/docs/.
- Editor agents: Cursor, Windsurf, Zed, Cline, Continue
- Claude family: Claude Desktop and Claude Code
- OpenAI Codex
- Registry: sh.marginal/mcp
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.
- 100 experiments/month on the free tier
- Four MCP tools: generate_variants, launch_test, get_results, recommend_next
- Listed in the MCP Registry as sh.marginal/mcp