Works with Claude, ChatGPT and other MCP apps

Keep track of anything by telling your AI.

Tell Claude, ChatGPT or another AI assistant what you ate, spent, lifted or read. It saves each entry in your own tables, and you get charts and totals for all of it. When you ask about last week, it looks up what you actually saved.

Add this URL where your assistant asks for a connector or MCP server, then sign in to MCP Cubby. Some apps, like the Codex app and Claude Code, wait for you to start the sign-in. Steps for each app

How it works

Say it. It’s filed.

  1. Connect once. Add the address above to Claude, ChatGPT or any app that supports MCP, then tap Start instantly. No sign-up first.
  2. Say what you want to track. “I want to track my workouts.” Your assistant suggests the columns, checks them with you, and sets it up.
  3. Then just talk. Every entry lands in the right table, and totals, goals and charts update straight away.
Your assistantMCP Cubby connected

3 sets of 5 squats at 100 kg

Saved to Fitness / Workouts

Chicken wrap and a flat white for lunch

Saved to Diet / Foods

Fitness bin 0
Diet bin 0
Sleep bin 0
Money bin 0
Mood bin 0
Reading bin 0
Asking

Ask. It’s found.

Without a record, an assistant can only go on what it remembers from your chats, so it guesses. With Cubby it looks the answer up: it goes to the right table, pulls out only the entries that match, and answers from those.

Your assistant

How much did I spend eating out last month?

Without a record

Maybe around $60? You mentioned a couple of lunches out.

Looked in Money / Spending

With Cubby

$175.40 on 6 meals out. The biggest was birthday dinner, $64.00 on 22 Oct.

Money / Spending

category = Eating out · 1 Oct – 31 Oct

  1. SupermarketGroceries$86.20
  2. Noodle barEating out$18.50
  3. Power billBills$142.00
  4. Corner caféEating out$12.40
  5. Train passTransport$45.00
  6. Pizza with SamEating out$36.00
  7. SupermarketGroceries$94.75
  8. Sushi lunchEating out$22.90
  9. Running shoesShopping$139.99
  10. Birthday dinnerEating out$64.00
  11. PharmacyHealth$17.30
  12. Fish and chipsEating out$21.60

6 of 12 entries matched · added up in code

The numbers add up

Code does the maths. Every time.

Totals, averages, streaks and goal progress are worked out from what you’ve saved, the same way every time. AI models are unreliable at arithmetic, so we don’t ask them to do it.

Volume = sets × reps × weightEstimated 1RM = weight × (1 + reps ÷ 30)

MCP CUBBY
FITNESS / WORKOUTS
THIS WEEK

9 NOV 2026NO. 0412

Mon Squat3 × 5 × 100 kg1,500

Mon Bench Press3 × 8 × 80 kg1,920

Wed Deadlift1 × 5 × 140 kg700

Wed Barbell Row3 × 10 × 60 kg1,800

Fri Squat3 × 5 × 102.5 kg1,538

Fri Overhead Press3 × 6 × 50 kg900

TOTAL VOLUME8,358 kg

BEST EST. 1RMDeadlift163.3 kg

SESSIONS3

CALCULATED IN CODE
NOT BY THE MODEL

Cubbies

Pull out a bin.

A cubby is one part of your life, like Fitness or Money. Inside are tables with typed columns: kilograms, hours and minutes, your currency. Some columns fill themselves in. Start from one of these, or describe your own. You can rename, add or remove anything later.

Fitness bin

Training sessions, body measurements and cardio.

Workouts7 columns
  • Exercisechoice
  • Setsnumber
  • Repsnumber
  • Weightweight
  • RPEnumber
  • Volume= sets × reps × weight
  • Est. 1RM= weight × (1 + reps ÷ 30)
Body3 columns
  • Weightweight
  • Body fatnumber · %
  • Waistnumber · cm
Cardio4 columns
  • Activitychoice
  • Distancedistance
  • Timeduration
  • Avg heart ratenumber · bpm
Diet bin

What you eat, with calories and macros.

Foods6 columns
  • Foodtext
  • Mealchoice
  • Caloriesnumber · kcal
  • Proteinnumber · g
  • Carbsnumber · g
  • Fatnumber · g
Water1 columns
  • Amountnumber · ml
Sleep bin

How long and how well you slept.

Nights3 columns
  • Time asleepduration
  • Qualityrating
  • Bedtimetext
Money bin

What you spend and earn, by category.

Spending3 columns
  • Amountmoney
  • Categorychoice
  • Merchanttext
Income2 columns
  • Amountmoney
  • Sourcechoice
Mood bin

How you’re feeling each day, plus a journal.

Check-ins3 columns
  • Moodrating
  • Energyrating
  • Feelingchoice
Journal2 columns
  • Entrytext
  • Topicchoice
Habits bin

Habits you want to keep up, and your streaks.

Check-ins2 columns
  • Habitchoice
  • Doneboolean
Reading bin

Books you want to read, are reading and have finished.

Books5 columns
  • Titletext
  • Authortext
  • Statuschoice
  • Ratingrating
  • Pagesnumber
Sessions3 columns
  • Bookchoice
  • Pagesnumber
  • Timeduration
Health bin

Symptoms and medication.

Symptoms2 columns
  • Symptomchoice
  • Severityrating
Medication2 columns
  • Medicationchoice
  • Dosetext
Work bin

Hours spent on each project.

Time4 columns
  • Projectchoice
  • Hoursduration
  • Billableboolean
  • Tasktext
+

Your garden, your car, a new baby, guitar practice, gift ideas.

Just describe it

“I want to keep track of my garden: what I planted, when I watered, and what came up.”

Your assistant proposes the tables and columns, checks them with you, and builds them. Every table gets its own dashboard.

Your data

Your house. Your keys.

Every app that connects gets its own key. See which ones have used your data and when, take any key back with one tap, and download everything as JSON whenever you like. Use ChatGPT on your phone and Claude on your laptop: both open the same cubbies.

Claude 2 min ago ChatGPT yesterday Cursor last week key taken back Old laptop 3 weeks ago
For developers

Building your own agent?

Your agent can create an account with one API call, no person needed. It gets a claim link to pass on to its user, who can take the account over whenever they’re ready. Read the agent guide

curl -X POST https://mcpcubby.com/api/agents/register

-H "Content-Type: application/json"

-d '{"agent_name": "My agent"}'

api_key · cub_k_••••••••••••

mcp_url · https://mcpcubby.com/mcp

claim_url · https://mcpcubby.com/claim/•••••

Claude ChatGPT Cursor Claude Code VS Code Your own agent

Make some room.

Connecting takes about a minute. 15 tools, one address, every assistant you use.