Report a Problem to LMCP From Your AI on Mac

The report_problem tool lets your AI assistant send a bug report, feature request, or integration request to the LMCP team without leaving the chat. Instead of hunting for a support form, you just describe what went wrong (or what you wish existed) and the AI files it for you. A typical prompt looks like: "Report a problem: list_emails is timing out on my iCloud account and returns nothing after 30 seconds." The AI captures your message, packages it, and sends it through report_problem so the team sees it directly.

It works just as well for ideas as for bugs. Try: "Send a feature request to LMCP — I'd love a tool that exports my Reminders to a CSV file." or "Request an integration with Linear so my AI can read my issues." This is a send-type tool, so it transmits your feedback to LMCP; nothing else on your Mac is touched. Download LMCP to get started.

Which AI agents work?

LMCP exposes report_problem to every major MCP-capable assistant:

• Desktop clients (auto-configure via local stdio): Claude Desktop, Cursor, VS Code (GitHub Copilot), Windsurf, and Zed detect LMCP automatically and run it over a local connection — no manual setup.
• Web AIs (via the LMCP Cloud Relay connector): ChatGPT, Claude.ai (web), Grok, and Perplexity connect to your Mac through the secure LMCP Cloud Relay, so you can file feedback from the browser too.

Automation

Because report_problem is one of 150+ tools in LMCP, it chains naturally with everything else. When a tool misbehaves, your AI can run run_diagnostics or get_config first, then fold those results into the report so the team gets real context instead of a vague "it broke." After a workflow — say the AI reads your calendar, drafts an email, and notices a parsing glitch — it can finish the job and then file a report about the glitch in the same turn. You can also ask it to bundle several small annoyances into a single, well-structured feature request.

Context

The AI runs against your real Mac data, so its reports are grounded in what actually happened. If search_messages returned an error on your machine, the AI already has the exact failure in front of it and can describe the symptom, the account, and the tool involved. That precision is what makes a feedback channel useful: the report reflects your genuine setup — your accounts, your apps, your version — not a generic guess.

Productivity

Filing good feedback usually means switching apps, finding the right form, copying error text, and explaining your environment. With report_problem the AI does all of that in one sentence of yours. A report that used to take five minutes of context-gathering becomes a single prompt. The faster the team hears about a broken tool or a missing integration, the faster it ships — and your one-line report directly shapes what gets built next.

Privacy & GDPR

LMCP runs entirely on your Mac. Your apps, accounts, and files are read locally and never copied to a server. When you use report_problem, only the message you choose to send reaches the team — there's no background harvesting of your data, no cloud mirror of your mailbox or contacts. This local-first design is private by default and GDPR-compliant by architecture. You control exactly what gets sent, every time.