What does lmcp_state actually return?

A structured snapshot of your LMCP environment: the server, tray, and teams-proxy versions, the detected AI client, whether the Cloud Relay connector is online, and the state of your TCC permissions for Calendar, Contacts, Reminders, and Automation.

Does lmcp_state change anything on my Mac?

No. It is a read-only tool. It only reports the current state of your LMCP setup and never modifies settings, permissions, or data.

Can I run lmcp_state from ChatGPT or Perplexity?

Yes. Web AIs like ChatGPT, Claude.ai, Grok, and Perplexity connect through the LMCP Cloud Relay connector, while desktop clients like Claude Desktop, Cursor, VS Code, Windsurf, and Zed auto-configure over local stdio.

Check LMCP Status & Health on Mac with AI

The lmcp_state tool returns a structured snapshot of your Local MCP environment on macOS. In one call it reports the server, tray, and teams-proxy versions, which AI client is talking to it, whether the Cloud Relay connector is online, and the state of your TCC permissions (Calendar, Contacts, Reminders, Automation). It is the fastest way to answer "is everything wired up correctly?" without digging through logs or System Settings.

A concrete prompt you can type into your AI assistant: "Check my LMCP state and tell me if anything is misconfigured." The assistant calls lmcp_state, reads the snapshot, and replies in plain language — for example, that your server is on the latest version but Contacts permission has not been granted yet, with the exact step to fix it.

Which AI agents work?

LMCP is a native macOS MCP server with 150+ tools, and lmcp_state works across every supported assistant:

Desktop clients (auto-configure via local stdio): Claude Desktop, Cursor, VS Code (GitHub Copilot), Windsurf, and Zed. LMCP writes the MCP config for these automatically, so the tool appears after a restart.
Web AIs (connect through the LMCP Cloud Relay connector): ChatGPT, Claude.ai (web), Grok, and Perplexity. These reach your Mac securely through the relay, so you can run lmcp_state from a browser tab too.

Automation

Because it is read-only, lmcp_state is the natural first step in any troubleshooting or onboarding chain. The AI can call it before running another tool to verify that the relevant permission is granted — for example, checking TCC state before list_calendar_events, or confirming the relay is up before a web AI queries your Mail. Pair it with run_diagnostics and get_config and the assistant can self-heal a broken setup: detect the gap, explain it, and tell you exactly which toggle to flip.

Context

Unlike a generic status page, lmcp_state reflects your real machine: the actual versions you have installed, the AI client currently connected, and the real permission grants on your Mac. The assistant works with this live context instead of guessing, so its advice is specific to your environment rather than boilerplate.

Productivity

Setup problems are the number one reason an AI assistant "can't see" your apps. Instead of trading screenshots back and forth or reading server logs, you ask one question and get a precise diagnosis in seconds. A check that used to take ten minutes of manual inspection across System Settings, version numbers, and config files becomes a single sentence. For teams onboarding new users, it turns "it doesn't work" into an actionable answer immediately.

Privacy & GDPR

Everything runs locally on your Mac. lmcp_state reads your environment on-device and returns the snapshot straight to your AI session — no copy is sent to or stored on any server, and there are no API keys or accounts to create. Your data never leaves your machine. This is privacy by design: read more about how LMCP is GDPR-compliant by architecture.

Download LMCP to give your AI assistant a live view of your Mac setup.