What file types can word_read open?

It reads modern Word documents in the .docx format. You can point to a file by full path (like ~/Documents/report.docx) or describe it by name and folder, and the tool returns the document's text content.

Does my Word file get uploaded to a server?

No. word_read runs entirely on your Mac through the local LMCP server. The text is read on-device and passed to your AI client. No copy of the file is stored in the cloud, and there are no API keys involved.

Which AI assistants can read my Word documents?

Claude Desktop, Cursor, VS Code (GitHub Copilot), Windsurf and Zed connect automatically over local stdio. ChatGPT, Claude.ai, Grok and Perplexity connect through the LMCP Cloud Relay connector.

Read Word Documents with AI on Mac (.docx)

The word_read tool lets an AI assistant open any Word document (.docx) on your Mac and read its text content. Instead of opening the file, copying paragraphs and pasting them into a chat window, you just ask. For example, you might type: "Read ~/Documents/Q3-report.docx and give me a three-bullet summary of the conclusions." The AI reads the file directly from disk and answers — the full text never leaves your machine.

It works on contracts, reports, meeting notes, drafts, proposals — any .docx file you can point to by path or describe by name. Download LMCP to enable it.

Which AI agents work?

LMCP exposes word_read to any MCP-capable assistant. Desktop clients auto-configure over a local stdio connection — nothing to paste:

Claude Desktop
Cursor
VS Code (GitHub Copilot)
Windsurf
Zed

Web-based AIs connect through the LMCP Cloud Relay connector, which bridges the browser to your local Mac server:

ChatGPT
Claude.ai (web)
Grok
Perplexity

Automation

Reading a document is rarely the end goal — it's the first step in a chain. Because word_read is one of 150+ tools in LMCP, the AI can string actions together in a single request:

Read a proposal, then draft an email reply summarizing it through the Mail tools.
Read meeting notes, then create reminders or calendar events for each action item.
Read several .docx drafts, then write a combined version back out with the Word writing tools.
Read a contract, then save key dates as calendar events so nothing slips.

You describe the outcome; the AI picks the tools and runs them in order.

Context

The AI works against your real files, not a generic example or a re-uploaded copy. When you ask it to compare last quarter's report with this one, it reads both actual documents sitting in your folders. That grounding means the answers reflect your real data — names, numbers and wording exactly as they appear on your disk, with no transcription errors from copy-paste.

Productivity

Skimming a 20-page Word document for the one clause you need can eat ten minutes. Asking "What's the termination notice period in this contract?" takes seconds, and the AI quotes the exact sentence. Multiply that across a folder of reports, proposals and notes and word_read turns document review from a chore into a quick conversation. No app switching, no scrolling, no manual copying.

Privacy & GDPR

Everything runs locally on your Mac. LMCP is a native macOS MCP server: the document text is read on-device and handed straight to your AI client. There are no API keys to manage, no cloud storage, and no copy of your file is ever uploaded to an LMCP server. This makes LMCP GDPR-compliant by architecture — your data simply never leaves the machine it already lives on. Read-only tools like word_read only access the specific files you reference in a prompt.

LMCP is free and native to macOS. Download LMCP and start asking your AI to read your documents.