
One JSON block in your project and Cursor's agent wires email while it builds the feature — sends the welcome mail, stands up the inbound webhook, tests the round trip.
Free sandbox to integrate · Production from $3/month
Cursor reads MCP servers from .cursor/mcp.json in the project (or globally from your home directory). Paste the block, reload, and the agent panel lists the Drin tools.
Full setup guide →{
"mcpServers": {
"drin": {
"command": "npx",
"args": ["-y", "@drin00/mcp"],
"env": { "DRIN_API_KEY": "drin_k_…" }
}
}
}
# project-level shown; ~/.cursor/mcp.json works globally.Three real tasks, end to end — copy a prompt and watch it run the tools.
Using the drin tools: create an inbox support@mydomain.com and a webhook pointing at our /api/inbound route, then simulate an inbound message and show the parsed JSON.
Send a test email from hello@mydomain.com to me with subject "Cursor wired up", then check delivery and summarise the event trail.
Draft a campaign to the NEWS topic announcing this release — use the merged PR titles for highlights — and route it to me for approval. Do not launch it.
Both work. Keep it per-project in .cursor/mcp.json when the key belongs to that product; use the global file if you want Drin in every workspace.
Use a scoped key, and if the repo is shared, reference an environment variable instead of committing the literal — Cursor resolves env values from your shell. Rotate keys from the dashboard any time.
Yes — that's the point. In agent mode Cursor plans multi-step tasks, so it can scaffold your email flow and verify it end to end with real tool calls.
Start with a sandbox key: sends are limited to verified recipients while you integrate. Campaigns and automations always stop at approve-by-reply — a human launches, not the agent. The guardrail →