macroline
← Blog

Why we built an MCP server for nutrition data

Macroline runs as a Model Context Protocol server. Here's what that means, why it matters, and how AI-native tools change what a tracker can be.

Alex Rivera Alex Rivera MCPAIdeveloper tools

The Model Context Protocol (MCP) lets AI assistants — Claude, ChatGPT, custom agents — read from and write to external systems through a standardized interface. Think of it as USB for AI tools.

Macroline runs as an MCP server. Once you connect it to Claude (or any MCP-compatible client), you can:

  • Log meals from any AI conversation. “I had pasta for lunch, log it” — Claude figures out the macros and writes it to your diary.
  • Query your own data conversationally. “What was my average protein intake last week?”
  • Build custom dashboards. Ask Claude to summarize your trends, generate weekly reports, plan meals around your remaining macros.
  • Integrate with anything. If it speaks MCP, it speaks to Macroline.

Why this is more than a chatbot integration

The lazy version of “AI macro tracking” is a ChatGPT wrapper that calls a nutrition API. We didn’t build that.

What we built is the inverse: the tracker is the source of truth, and AI clients are dispatchers. Your diary, your goals, your weight, your saved meals — they all live in Macroline. The AI just talks to them.

This matters because:

You own your data. Switch AI clients tomorrow — the data doesn’t move. We’re not lock-in software.

The food database is the moat. 8,500+ USDA foods, OpenFoodFacts barcodes, chain restaurants researched on demand, every row sourced. AI clients all benefit from the same authoritative database.

Privacy is bounded. AI clients only see what your account allows. You don’t dump your whole life into the model — you grant scoped access through OAuth.

What you can do today

The MCP server is in active development. The first wave of tools:

  • search_food — Find a food by name or brand
  • log_meal — Add diary entries with idempotency
  • get_diary — Read recent entries
  • get_summary — Daily macros + remaining
  • set_goals — Update calorie/macro targets

Future waves will add the research agent (chain restaurant lookups), barcode scanning via shared cache, and weight/medication tracking endpoints.

How to connect

Once we ship MCP OAuth, connecting Claude to Macroline will be a one-tap flow in the iOS app. Coming with v0.5.0.

Until then, the MCP server is being scaffolded internally. If you’re an AI developer who wants early API access, drop us a line.

The bigger bet

We think most apps will eventually be AI-accessible by default. The question is whether they’re AI-first (built around the assumption that an AI agent will be a primary user) or AI-bolted-on (a chatbot button stuck on a 2018 app).

Macroline is the former. The MCP server isn’t a feature — it’s how we think the next decade of consumer software should be built.