Overview

What Gutlog is, its concepts and every way to use it.

Gutlog is a food and gut diary your AI assistant writes to. Share a meal in ChatGPT or Claude and the assistant sends Gutlog a draft; Gutlog matches it against the food database, the barcode index and your own foods, stores calories, macros, nutrients and ingredients with a confidence for each, and keeps it ready for your assistant to read, audit and fix. The fastest start is the setup page.

Concepts

  • Meal: what you ate at one time, with its items. Always stored, even when unsure; uncertain meals are flagged needs_review.
  • Item: one food or drink in a meal, matched to a food or a barcoded product, with grams and nutrients.
  • Ingredient: what an item contains, with a confidence and a tier (confirmed, likely, possible). See meal drafts and confidence.
  • Nutrients: integers in a base unit per nutrient (millikilocalories, milligrams, micrograms). See nutrients and units.
  • Clean verdict, tummy score, trigger, program: the clean-ingredient rubric, your daily gut score, suspected trigger foods and elimination diet programs. They arrive as their ops ship; the API reference lists what is live.
  • Op: one action with a name and an input schema, such as log_meal or get_day. The web app, the MCP server, the REST API and the CLI all call the same ops.

Gutlog is not medical advice. Gut insights are hypotheses to test, not diagnoses.

Ways to use it

  • MCP: https://getgutlog.com/api/mcp, one tool per op, for ChatGPT, Claude, Claude Code, Codex, Cursor and other clients.
  • REST: POST https://getgutlog.com/api/v1/{op} with a JSON body.
  • CLI: npx @meterapp/gutlog, for terminals and scripts.
  • TypeScript SDK: a typed client for every op.
List every op (no key needed)
curl https://getgutlog.com/api/v1

Guides

Getting started
Connect ChatGPT or Claude with one prompt or the setup page, log a first meal, see what is left to set up and what is free.
Quickstart
Connect your assistant over MCP, log a first meal from a draft and read the day back, in five minutes.
Logging meals
Photos, sentences and barcodes; confidence tiers, the review queue and reading your diary back.
Fixing a meal
Fix anything in one sentence or one tap; your edits always win, and every change is kept.
Clean verdicts and organic
Our own ingredient rubric with an evidence level per flag, profiles, swaps and when organic is worth it.
Gut, triggers and programs
Tummy scores, suspected triggers as hypotheses, and low-FODMAP and single-food programs with a safety screen.
Data, audits and exports
Provenance on every row, audit_diary, and export_data as JSON and CSV with a data dictionary.
Privacy
What happens to your food and symptom data: never sold, never sent to analytics or ads, yours to export or delete.
MCP server
Connect ChatGPT, Claude, Claude Code, Codex, Cursor and any MCP client to Gutlog.
REST API reference
Every op, its input schema and a curl example, generated from the ops registry.
CLI
Run any Gutlog op from a terminal, a script or CI.
TypeScript SDK
A typed client for every op, with retries, idempotency keys and typed errors.
Errors
RFC 9457 problem responses and every error code the API returns.
Meal drafts and confidence
The draft an assistant sends to log_meal, ingredient confidence tiers (0.85, 0.60, 0.30) and when a meal needs review.
Nutrients and units
Every nutrient and its integer unit: energy in millikilocalories, macros and most minerals in milligrams, some vitamins in micrograms.
Accounts, roles and keys
Your personal organization, roles, API keys and their scopes.
Feedback and reviews
Request features, report issues with the failing request attached, upvote, comment and review, from the board, MCP, the CLI or REST.