Where does save_attachment save the file?

Wherever you tell the AI to put it — for example your Downloads folder or a specific project directory. If you don't specify a path, just ask the assistant to save it somewhere and it will write the copy to disk on your Mac.

Why should I mention the account or mailbox?

Passing account= (and mailbox= when known) lets LMCP jump straight to the right message instead of scanning every mailbox. It makes the attachment lookup much faster, especially if you have several accounts or a large inbox.

Does my email or attachment get uploaded anywhere?

No. save_attachment reads the attachment from Mail.app and writes it locally on your Mac. Nothing is sent to an LMCP server and no copy is stored in the cloud.

Save Email Attachments with AI on Mac — LMCP save

The save_attachment tool lets your AI assistant pull an attachment out of an email in Mail.app and save it to a folder on your Mac. Instead of opening the message, clicking the paperclip, choosing a destination and confirming, you just ask. A typical prompt looks like: "Save the PDF attached to the invoice email from Acme into my Downloads folder." The AI finds the message, grabs the file and writes it to disk for you.

It is a read-only tool over your mailbox: it reads the attachment and saves a copy where you ask, without altering or sending anything. For speed, the AI passes account= (and mailbox= when it already knows the folder) so the lookup is fast even on large mailboxes. LMCP ships 150+ tools, and save_attachment is one of the Mail tools that makes triaging your inbox conversational. Download LMCP to enable it.

Which AI agents work?

LMCP works with the AI assistant you already use. Desktop clients auto-configure over a local stdio connection, while web-based AIs connect through the LMCP Cloud Relay connector.

Desktop (auto-configure via local stdio): Claude Desktop, Cursor, VS Code (GitHub Copilot), Windsurf and Zed.
Web (via LMCP Cloud Relay connector): ChatGPT, Claude.ai (web), Grok and Perplexity.

Whichever you pick, save_attachment runs on your Mac against your real Mail.app accounts.

Automation

The real power shows when save_attachment chains with other LMCP tools. The AI can search_emails to find every message from a vendor this month, read_email to confirm which one carries the contract, then save_attachment to drop each file into a project folder. From there it can hand the saved path to the Office document tools to read a Word or Excel file, or use Finder tools to rename and organize the downloads. One sentence becomes a multi-step workflow: "Save all the receipts attached to emails from this week into ~/Documents/Receipts."

Context

Because LMCP talks to your real Mail.app, the AI works with your actual accounts, mailboxes and messages — not a sandbox or a paste of text. It knows which account an email lives in, which folder, and which attachment you mean. That grounding is why passing account= and mailbox= keeps things fast and accurate: the assistant operates on the exact data on your machine, the same files you would see if you opened Mail yourself.

Productivity

Saving attachments by hand is a small task you repeat dozens of times a week: open the message, find the paperclip, pick a folder, confirm, repeat. Batching it through your AI turns minutes of clicking into a single request. Accountants pulling invoices, lawyers collecting signed contracts, and anyone assembling files for a report all save real time when the assistant can grab and file attachments while they keep working.

Privacy & GDPR

Everything runs locally on your Mac. The attachment is read from Mail.app and written straight to your disk — it never travels to an LMCP server, and no copy of your file or email is stored in the cloud. There are no API keys to hand over and no third-party data processor in the path. This local-first design makes LMCP GDPR-compliant by architecture: your data simply never leaves the machine it already lives on.