Does it search hidden or background tabs too?

Yes. It scans the rendered text of every open Safari tab, including ones you are not currently looking at, and returns each tab that matches with a snippet of surrounding text.

Does this send my browsing data to the cloud?

No. safari_search_tabs runs entirely on your Mac. The matching tabs and snippets go directly to your AI session and are never stored on an LMCP server or copied to the cloud.

Do I need API keys or a Safari extension?

Neither. LMCP talks to Safari natively on macOS. You only install the LMCP app once and grant access; no keys, no extensions, no accounts.

Search Across All Open Safari Tabs with AI on Mac

The safari_search_tabs tool lets your AI assistant search the rendered text of every open Safari tab on your Mac for a word or phrase. It returns each matching tab along with the surrounding snippet, so you can instantly find which of your 30 open tabs actually mentioned that thing. You just type something like "Search my open Safari tabs for the pricing table that mentioned 'enterprise tier'" and the AI scans them all and tells you exactly where it lives.

This is the answer to tab overload. Instead of clicking through dozens of tabs trying to remember which article had the quote, the config snippet, or the flight price, the AI reads them for you and points to the exact match with context. It is read-only, so nothing on the pages is changed.

Download LMCP to add this and 150+ other native Mac tools to your AI assistant.

Which AI agents work?

LMCP is a free, native macOS MCP server, so any MCP-capable AI can use this tool.

Desktop clients — Claude Desktop, Cursor, VS Code (GitHub Copilot), Windsurf and Zed auto-configure over a local stdio connection. No setup beyond installing LMCP.
Web AIs — ChatGPT, Claude.ai, Grok and Perplexity connect through the LMCP Cloud Relay connector, which securely bridges the web client to the server running on your Mac.

Automation

safari_search_tabs becomes powerful when chained with other LMCP tools. The AI can find the matching tab, then read its full page text, summarize it into a note with the Notes tool, or drop the key quote into an email draft. A typical flow: search every tab for "Q3 roadmap", pull the relevant tab's content, and create a Reminder to follow up — all in one prompt, no copy-paste.

Context

The AI is not guessing from training data — it works with your real, live browser session. It sees the actual tabs you have open right now, the real text those pages rendered, and the actual snippets around your search term. That grounding means answers reflect what you were genuinely researching, not a generic web result.

Productivity

Anyone who keeps a wall of open tabs — researchers, developers, shoppers comparing options — loses minutes hunting for "the tab that said X." safari_search_tabs collapses that to one sentence. Find the doc page with the exact API parameter, the listing with the right price, or the article with that statistic, in seconds instead of clicking through twenty tabs.

Privacy & GDPR

Everything runs locally on your Mac. The tool reads your open Safari tabs in place and hands the result straight to your AI session — no page content is uploaded to an LMCP server, copied to the cloud, or stored anywhere. There are no API keys and no third-party accounts. This is GDPR-compliant by architecture: data the AI sees never leaves the device.