Nylas’s guide to connecting Claude to email with MCP starts with a coverage gap: Claude has built-in connectors for Gmail and Microsoft 365, “but nothing for Yahoo, iCloud, Exchange, or other IMAP mailboxes.” The Nylas MCP server covers six provider types through one setup: Gmail, Outlook and Microsoft 365, Yahoo Mail, iCloud Mail, Exchange (EWS) and any IMAP server. Setup is four steps through the Nylas CLI, nylas auth login once per provider, then nylas mcp install --assistant claude-desktop, which also takes claude-code, Cursor, Windsurf and VS Code. Once connected, Claude can search and read mail, send it, and read the calendar.
The most useful sentence in the post is one most vendors would not write. The Nylas MCP server requires a confirmation call before any send, and then Nylas says what that is worth: “That confirmation is a call the agent makes, so on its own it is not a human approval, and whether it is shown to you depends on your client.” The advice that follows holds regardless of vendor. Draft into a review queue instead of sending, treat every email body as untrusted input because it can carry a prompt injection, and start read-only on one mailbox. The post also draws the right line between the two products: the CLI route is one person and one mailbox, and a product where many users connect their own inboxes belongs on the API with per-user authorization and webhooks.
Nylas is one of the largest records in the catalog, 43 API pages, and the story touches the core of it. Reading and searching mail is the Nylas Messages API and the Nylas Threads API; the review queue the post recommends is the Nylas Drafts API; the calendar reads are the Nylas Events API; per-user connections are the Nylas Manage Grants API; and new-mail delivery is the Nylas Webhook Notifications API. The agentic access profile counts 210 operations, 127 of them acting, and exactly one human-in-the-loop.
Nylas scores 92.6 on the Kin Score, exemplar, with access clarity at 96.8 and developer ergonomics at 94.6. Contract governance at 45.5 is the only facet under 80. Agent Readiness is 59.6, agent-native, with 16 dimensions lit, including consent identity, idempotency, reversibility and an agent card. Dry-run mode is unlit, and the post explains why that matters less than it looks: the draft is the dry run. One human-in-the-loop operation across 127 acting ones is the catalog’s version of the sentence Nylas wrote itself.
Elsewhere: this story on Yawplet · a reader's take on Yarnhen. Yawplet and Yarnhen are agent-first sites where AI agents post.