Gym Snap Coach

Live

Photograph any gym, learn each machine, and get a routine from what is there.

musemuse-gym-snap-coach.vercel.appOpen

Get it

Connect it to any app that speaks MCP. It has 3 tools: detect_gym_equipment, explain_machine_use, build_workout_routine. No sign-in.

Muse app: add this URL as a custom connector.
Claude Code: claude mcp add --transport http gym-snap-coach https://muse-muse.vercel.app/api/mcp/gym-snap-coach
Anything else: add it as a remote server over streamable HTTP.

What was checked

Checked by code

Each line is what the check ran on, and no more.

Passed Typecheck · Nov 1, 2026
Passed Build · Nov 1, 2026
Passed The built app starts · Nov 1, 2026
Passed No errors in the browser console · Nov 1, 2026
Passed Smoke test of the main flow · Nov 1, 2026
Passed Its manifest is valid · Nov 1, 2026

Judged by a model

This is a model's opinion, not proof.

Passed A model's opinion of the screenshots, against the brief · Nov 1, 2026

Not checked

Nothing here should be taken as tested.

Not checked Whether its answers are true
Not checked ChatGPT: only a person with an account can test it
Not checked Installing the Claude plugin or the Muse gadget pack on a real device
Not checked No Prove it run: nothing it comes with was run for real

The muse

A short text note asking for a tool that photographs a gym, segments and crops the machines, explains how to use each one, and builds a workout from the equipment found.

Took from it

The core job: turn one gym photo into machine guidance and a workout routine.
The flow: detect equipment first, then explain one machine, then build a plan from what is available.
A practical and encouraging tone for beginners who feel unsure in a gym.

Left out

Models

SAM 3.1Find and outline each machine in the gym photo so the app can crop and label individual equipment
Muse WonderIdentify each cropped machine, explain setup, form and safety, and build an ordered workout routine from the machines found
Muse ImageCreate a clean how-to illustration of correct setup and form for the selected machine

Cost per run: $0.05 est.
Runs so far: 213 calls · 228.7k tokens · $3.18
Build cost: $8.58 est.

Limits

Machine identification from a photo can be wrong, especially for similar cable or selectorized machines. Check the name and weight stack on the machine itself before use.
Generated illustrations show general form only. They are not a substitute for a qualified trainer, and joint angles or grips may not match your exact machine model.
Only equipment clearly visible in the photo is included. Machines hidden, cut off, or in shadow may be missed or cropped poorly.
The routine is general fitness guidance, not medical advice. Stop if you feel pain and talk to a professional about injuries or health conditions.
One routine and one illustration are generated per run to keep use fast and low cost.

Muse app connector

Add it in the Muse app as a custom connector. It exposes 3 actions: detect_gym_equipment, explain_machine_use, build_workout_routine.

Recorded demo

How it was built

Muse Code made 102 tool calls and 13 edits across 13 files (+867 −69).
npm run verify passed after 3 fix rounds.
Screenshot judge: pass. Final screen shows a complete Beginner Strength Session in 40 Minutes with warm up, ordered exercises, and cool down.
Built in 40 min for about $8.58 in model calls (est.).
Muse Code was stopped after 12 minutes.
How it was built: 102 tool calls, 13 edits touching 13 files { <read-skill-result name="meta-ui" status="ok"> { Edited `app/api/actions/detect_gym_equipment/route.ts` Edited `app/api/actions/explain_machine_use/route.ts` Edited `app/api/actions/build_workout_routine/route.ts` Edited `app/page.tsx` { Edited `musemuse.json` Edited `tests/smoke.spec.ts` Edited `tests/actions.spec.ts` Edited `tests/errors.spec.ts` Edited `BUILD_NOTES.md` { { (failed) { Read image file `test-results/judge/02-machines-found.png` as model-visible image output. Read image file `test-results/judge/03-machine-guide.png` as model-visible image output. Read image file `test-results/judge/01-start.png` as model-visible image output.