pylos-mcp

Read-focused, prompt-injection-hardened email MCP server for any IMAP provider.

Anyone in the world can put text in your inbox, and the moment an AI assistant reads that inbox, anyone in the world can put text in front of your assistant. pylos-mcp is an email MCP server built around that fact. It lets Claude, or any MCP client, search, read and draft your mail while treating every message as what it really is, input from a stranger. Mailbox content is fenced as data before the model sees it, and there is no bcc field for an injected email to quietly copy someone with.

It runs on your machine and speaks plain IMAP, so it works with Gmail, iCloud, Yahoo, GMX, Fastmail, mailbox.org, Posteo, Proton via Bridge, or anything self-hosted, and your credentials never leave home. Out of the box it can read and draft. Anything riskier, moving, sending, deleting, is a separate switch that stays off until you flip it.

What this can never do

Mail is attacker-controlled text, so the hard limits live in the architecture rather than in a prompt. No message can talk the server out of any of these.

• No raw HTML ever reaches the model. Bodies come from the plain-text part when one exists or are converted to text otherwise, and invisible characters that could hide instructions from a human reader while staying readable to a model are stripped along the way.
• Untrusted content is fenced. Everything that came from a mailbox, bodies, subjects, sender names, folder listings, Sieve script text, is wrapped in a labeled delimiter before the model sees it, and the delimiter sequence is neutralized wherever it appears inside the content, so a malicious email cannot forge a closing marker and write instructions outside the fence. The few lines that live outside the fence (the metadata line above a message, the one-sentence confirmations a tool returns after it acts, the Warnings line) are server-authored, stripped of hidden characters, collapsed onto a single line and length-capped, and the Warnings line never carries message content at all.
• No bcc field exists anywhere, not on drafts and not on sent mail. A bcc recipient receives a full copy of a message while appearing nowhere in it, and that invisibility is exactly what an injected email would want, a silent extra recipient that no review of the draft or of the sent copy could ever catch. The field is absent rather than guarded, so there is nothing to talk the model into.
• Deleting a message moves it to Trash. There is no expunge and no permanent-delete option, and the tool result never claims a permanence this server does not offer.
• Sieve access is read-only, permanently. Server-side filter rules can forward, auto-reply and notify, each an exfiltration channel that survives revoking the app password or uninstalling this server. Scanning uploaded scripts for dangerous commands would only be safe if this project's parser agreed with the mail server's parser exactly, and any disagreement between the two is a bypass, so write access is left out entirely rather than defended.

Sending is the other risky door, so it starts closed even once the send capability is on. Until SEND_ALLOWLIST says who may be addressed, every send is refused, and the refusal names the two ways to open the gate. Choosing SEND_ALLOWLIST=* allows anyone, visibly and on purpose.

Fencing reduces prompt-injection risk, nothing eliminates it. The model still reads text written by strangers, so treat every response that includes message content as untrusted input rather than ground truth.

Quick start

Add the server to your MCP client's config. For Claude Desktop that file is claude_desktop_config.json.

{
  "mcpServers": {
    "pylos-mcp": {
      "command": "npx",
      "args": ["-y",

... [View full README on GitHub](https://github.com/adamvass/pylos-mcp#readme)