CAN-SPAM Compliant Test Sends with marginal
If you're testing subject lines from an AI agent, your test sends still need to follow the same rules as any commercial email. Here's how to keep CAN-SPAM compliant test sends front of mind while you experiment through marginal's email MCP server.
What CAN-SPAM asks of every send
CAN-SPAM applies to commercial email regardless of how small the batch or how experimental the goal. A subject line A/B test is still a real send to real inboxes, so the same baseline obligations apply to each variant you launch.
- Don't use deceptive subject lines — the line you're testing must reflect the actual message content.
- Include a valid physical postal address in the email body.
- Provide a clear, working unsubscribe mechanism and honor opt-outs promptly.
- Identify the message as an advertisement where required, and use accurate From and reply-to details.
Where marginal fits — and what stays your responsibility
marginal is a hosted email MCP server that lets an AI agent generate variants, launch tests, and read results. It handles managed sends and open/click tracking, while compliance content — your physical address, unsubscribe link, and accurate sender identity — lives in the campaign content and list you bring to the test.
Because deceptive subject lines are a specific CAN-SPAM concern, keep your generated variants honest. When you prompt generate_variants, anchor it to the true offer so every A/B test option stays accurate as well as high-performing.
- generate_variants — produce subject line candidates that match the email's real content.
- launch_test — send the A/B test to your audience with tracking attached.
- get_results — review opens, clicks, and within-test lift.
- recommend_next — get a data-backed suggestion for the follow-up send.
A compliant testing workflow for AI agent email marketing
The goal is to make CAN-SPAM compliant test sends a default, not an afterthought. Build the checks into your template and your agent's instructions so nothing slips through when you move fast.
- Bake the unsubscribe link and postal address into the template before any test launches.
- Send only to recipients who opted in; never use test mode to email cold or purchased lists.
- Have your AI agent verify subject-line variants describe the body accurately before launch_test.
- Keep suppression and opt-out handling current so honored unsubscribes never receive a follow-up variant.
Connect and start testing
marginal connects to agent clients like Cursor, Claude Desktop, Claude Code, Windsurf, Cline, Continue, Zed, and OpenAI Codex. Point your client at https://marginal.sh/mcp, authenticate with your Bearer API key, and run subject line A/B tests through the MCP tools.
The free tier covers 100 experiments per month, which is plenty to validate a compliant testing process. See the docs at https://marginal.sh/docs/ for tool details and setup.
- Endpoint: https://marginal.sh/mcp
- Auth: Bearer API key (marg_live_...)
- Free tier: 100 experiments/month
- Tools: generate_variants, launch_test, get_results, recommend_next
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