Meals in plain language
Say “oatmeal with banana and peanut butter” — your AI estimates calories and macros, fiber, total sugar and caffeine included, and logs it. No database search.
Cronometer alternative
Cronometer doesn't have one — so you can't use it inside Claude or ChatGPT. Nutrition MCP does the same job by conversation, and it's free and open source.
The short answer
The Model Context Protocol (MCP) is the open standard that lets AI assistants like Claude and ChatGPT connect to outside tools. Cronometer doesn't publish an MCP server, so there's no official way to log food to it from your AI. If you searched for “Cronometer MCP” or “connect Cronometer to Claude,” what you're really after is a nutrition tracker that lives inside your AI — that's exactly what Nutrition MCP is.
What you get instead
Say “oatmeal with banana and peanut butter” — your AI estimates calories and macros, fiber, total sugar and caffeine included, and logs it. No database search.
Send a product barcode and pull the label macros from Open Food Facts — fiber and sugar too, where the label lists them. No Premium subscription to unlock it.
Log body weight in kg or lb, set calorie, macro, fiber, sugar, caffeine, and water goals — fiber a target to reach, sugar and caffeine limits to stay under — and track trends toward a goal weight. Alcohol tracking is there too, opt-in and off unless you turn it on.
Ask for daily totals, weekly trends, streaks, and recurring meal patterns — right in the chat.
Import your history from another app's CSV export — parsed in your browser, not by the AI. Export it all back out, or delete your account, whenever you want.
MIT-licensed and self-hostable — no ads, no paywall, no upsell. Audit the code or run your own instance.
Cronometer vs. Nutrition MCP
Cronometer is excellent if you want deep micronutrient precision. Nutrition MCP takes a lighter, conversational approach to calories, macros, and weight — right inside your AI.
Moving from Cronometer
Cronometer earned its reputation on precision — curated databases and tracking for 80+ micronutrients, vitamins and minerals included. If that micronutrient depth is why you open it, be honest with yourself: conversational estimates won't match a lab-grade database entry gram for gram.
But most people log to keep calories and macros in range, not to audit their selenium intake. That range is wider than it sounds: alongside protein, carbs, and fat you get fiber, total sugar, and caffeine in milligrams, and optional alcohol in grams of ethanol if you switch it on. For that, describing a meal to your AI is far less work than searching for and weighing every component — and you still get daily totals, trends, and a target weight to track against, for free.
There's also a middle path: because you're inside an AI assistant, you can ask for the micronutrient angle when you actually want it — “roughly how much iron and B12 was in today's meals?” — and get a reasoned estimate on demand, without the overhead of logging every gram to a curated entry the rest of the time.
Your Cronometer history
Ask to import and an importer opens right in the chat: pick your export, map the columns, preview what will be added, then confirm. The file is read in your browser — the AI never sees the rows. In clients without in-chat panels, paste your export instead.
Precision is why you used Cronometer, so a sloppy import would be worse than none. Ask to import and a panel opens in the chat: you pick your Cronometer CSV, it's parsed in your browser, and you approve a preview before a single row is written. The numbers are read straight out of the file — the AI never sees the rows, so it can't round or retype one.
Cronometer's export shape is recognised by name. It splits the timestamp across separate date and time columns, and both are read, so a breakfast logged at 07:12 keeps its time instead of landing at a default midday. It writes a quantity with the unit inside the same cell — “58.00 g”, “1.00 cup” — and a value written that way still reads as the number it is rather than as nothing. And it repeats the “Amount” heading more than once, so columns are keyed by position rather than by name: the duplicates can't silently collide, and the mapper tells you which one you're pointing at.
Be clear about what crosses over: the date and time, food name, meal, calories, protein, carbs, fat, fiber, total sugar, caffeine, and notes. Cronometer is the one export in this list that ships a Caffeine (mg) column, and it lands as milligrams — the unit it is already in, and the one caffeine is stored in here, so nothing is converted. A caffeine column headed in grams is left unmapped instead, with the reason shown, rather than recording 0.18 where the label says 180 mg. Sugar means total sugars, the fruit and milk included — not added sugar, which no export reliably carries. Cronometer's separate “Sugar Alcohols” column is a polyol rather than a sugar or an ethanol, and it can't land in either field. Alcohol is a special case: Cronometer exports it as ethyl alcohol in grams, and it comes across only if you've turned alcohol tracking on here first, since it's off until you do. Portion amounts and Cronometer's 80-plus vitamins and minerals don't cross over at all — that micronutrient depth stays in Cronometer's own export. Re-importing is harmless: each row carries a content fingerprint, so a second run of the same file reports the meals as already logged rather than adding them twice.
How to switch
Works with any MCP client that supports OAuth 2.0 with PKCE. On first connect you create an account with Google or an email and password.
https://your-domain.com/mcp
into the
Remote MCP server URL field and
click Add.
Using ChatGPT or another client instead? The full install guide covers ChatGPT, Cursor, VS Code, Claude Code, and more.
FAQ
No. Cronometer does not offer a Model Context Protocol (MCP) server, so there is no official way to connect it to Claude, ChatGPT, or other AI assistants. Nutrition MCP is a free, open-source alternative built as an MCP server from the ground up, so you can log meals and macros directly inside your AI.
There is no official Cronometer connector for Claude, because Cronometer has no MCP server or public MCP integration. The closest option is Nutrition MCP, a free MCP server: add https://your-domain.com/mcp as a custom connector in Claude, sign in, and start logging by conversation.
No. Cronometer's tracking of 80+ vitamins and minerals is its specialty, and Nutrition MCP has no micronutrient data at all — no sodium, no vitamins. What it does track is calories, protein, carbs, fat, fiber, total sugar, caffeine in milligrams, optional alcohol, water, and weight. You can still ask your AI for a rough micronutrient read on a meal, but if lab-grade micronutrient depth is essential, Cronometer is the better fit.
For calories, macros, fiber, and sugar, conversational estimates are close enough for most goals — but they won't match Cronometer's curated, gram-for-gram database. It trades a little precision for far less logging effort, which is the right trade for most people.
If you want to track calories, macros — fiber, total sugar, and caffeine included — water, and weight without opening a separate app or searching a food database, yes. Instead of tapping through a database, you describe what you ate in plain language, send a photo, or scan a barcode, and your AI logs it — completely free and open source.
Yes. Ask to import and an importer opens in the chat: you choose your Cronometer CSV, it's parsed in your browser rather than read by the AI, and you preview what will be added before confirming. Cronometer's export is recognised by name — its separate date and time columns are both read, and its repeated “Amount” heading can't collide because columns are keyed by position. The date and time, food name, meal, calories, protein, carbs, fat, fiber, total sugar, caffeine in milligrams, and notes come across; alcohol does too, but only if you've switched alcohol tracking on first. Vitamins, minerals, and portion amounts don't. Re-importing the same file never creates duplicates. In clients without in-chat panels you can paste your export instead.
Not when the importer opens. It parses the CSV in your browser and shows you what will be added before anything is written: how many meals, the calorie total, anything it had to flag, and the rows themselves — a long file lists the first of them plus a count of the rest rather than every line. Only the rows you confirm are sent, and they go as structured data rather than through the AI's reply, so no row can be mistyped or invented in transit. Each row also carries a content fingerprint, so running the same file again reports those meals as already logged instead of duplicating them. If your client can't display in-chat panels, the fallback is to paste the export — the AI does read it on that path, so prefer the importer when you have the choice.
Yes. Nutrition MCP is completely free with no premium tier, ads, or paywalled features — unlike apps that put some features behind a subscription. You only need a Claude or ChatGPT account to connect.
Free and open source — no Cronometer account, no app to open.
Cronometer is a trademark of its respective owner. Nutrition MCP is an independent, open-source project and is not affiliated with, endorsed by, or sponsored by Cronometer. Comparisons reflect publicly available information at the time of writing and may change.