MCP server · Analytics
LLMxRay MCP server
by LogneBudo
Ask your local AI what went wrong with its own Ollama calls: cache loss, context overflow, slow turns.
LLMxRay is a small helper you run on your own computer that watches what your local AI is doing. It sits between your chat tools and Ollama, the program that runs AI models on your machine, and records each turn. It is handy for anyone who runs a local model and wonders why it is slow, why it forgets things, or why it suddenly stopped using the fast path.
What is an MCP server? The 30-second version
On its own, your AI can only chat with you. An MCP server is a small helper program that gives your AI a new skill or a connection to something else. This one connects your AI to LLMxRay, the little monitor running on your machine, so the AI can look up its own past turns and answer questions about them. You ask in plain words, and the AI goes and checks the numbers for you.
What this MCP server does
You ask your AI something like why was my last answer so slow. The AI uses this helper to read the records that LLMxRay saved about your recent turns. The helper looks at token counts, cache reuse, speed and whether the model fit in video memory. Then the AI tells you in plain sentences what stood out and what you could change. Everything stays on your computer.
What you can do with it
- Ask what went wrong in your recent turns and get plain sentences with fixes
- Read your last turn in numbers: prompt size, cached share, speed, stop reason
- List which local models you have and whether each fits in video memory
- Compare turns before and after a change to see if it actually helped
- Check before sending whether a prompt of a given size will fit
- See the history of your turns, one line each, with flags
- Check where LLMxRay is running and how many turns it keeps
Try asking your AI
- “Why was my last turn so slow?”
- “Did my prompt cache break on the last few turns?”
- “Which of my local models fit in video memory right now?”
- “Compare my turns before and after I changed the system prompt”
What it gives back to you
You get back short answers in the chat, usually a few sentences plus a small list of numbers. For example, a diagnosis might say the prompt cache was lost on every turn and suggest moving the changing part to the end. A model list comes back as lines with size, whether it is loaded, and whether it fits in video memory. Nothing is changed on your machine; it only reads.
Before you start
What you need
- Ollama installed and running with at least one model pulled
- LLMxRay itself running (start it with npx llmxray)
- Node.js 18 or newer, if you use npx
- Your AI tool pointed at LLMxRay's port 11435 instead of Ollama's 11434
Good to know
It reads records of your AI turns, and if you switch on keeping the prompt text, that text is written to a local file, so be careful what you leave on.
Install it with your AI
Add LLMxRay MCP server to your AI, no technical skills needed
You don't install anything by hand. You copy one prompt, paste it into an AI that can work on your computer, and it checks, installs and connects the server for you, asking you when it needs something.
Sign in to get the install prompt
Members get a ready-made prompt that lets the Claude desktop app check LLMxRay MCP server, install it and connect it for them, step by step. You don't need any technical skills: you copy, paste and answer a few questions. Your connected AI can also find and install any of the 4,609 MCP servers here for you.
Who it's for
People who run local AI models on their own computer and want to understand why they are slow, forgetful, or behaving oddly.
