What it is, in one line.
A hosted, remote MCP server at https://api.myoamigo.com/mcp. You don't install or run anything — you point the AI assistant you already use at the endpoint, approve it once, and it can work with your training data on your behalf.
Lifters who already use AI
If you ask Claude or ChatGPT for training advice, this gives it your real numbers instead of guesses — PRs, recent sets, volume, what's stalling.
Reads, analyzes, and writes
Not a read-only mirror. Your assistant can draft a routine or a full weekly plan and it lands in the app, scheduled and ready to run.
Scoped to your sign-in
Every call is gated by your account. Nothing is shared by default, nothing trains a model, and you can disconnect anytime.
Things to ask once it's connected.
These aren't canned demos — they're ordinary questions your assistant can now answer against your actual log.
What should I do for bench today, based on my last few sessions?
Why has my squat stalled? Look at the last eight weeks.
Which muscles did I undertrain this month?
Build next week's workouts from my current program and put them on my calendar.
I just benched 185 for 5, 5, 4 — log it.
Compare my deadlift progress to where I was two months ago.
All 34 tools.
This is the live production surface, generated from the same source as tools.json, so it can't drift. Every call is scoped to the authenticated account.
Free to read, Pro to write. Every history & analysis tool below is free — connect your assistant and it can read your training, spot stalls, and draft a plan at no cost. The tools that change your data — logging sets, creating & editing routines and weekly plans — are part of MyoAmigo Pro. (Sending feedback and setting your profile stay free.)
History & analysis — 18 read tools
- get_user_context — dense snapshot: recent workouts, your exact exercise vocabulary, and per-lift e1RM / working ranges / PRs.
- get_history — recent logged sets for one exercise, most recent first.
- get_pr — personal record for an exercise: heaviest working set and best Epley e1RM.
- list_workouts — completed workouts with full per-exercise set detail (weight × reps, RIR, warmups).
- get_stats — aggregate stats: per-exercise e1RM / volume, volume by muscle, weekly trend.
- get_progress — recent PRs, stalls (lifts whose e1RM peaked sessions ago), and per-exercise trend.
- get_muscle_balance — per-muscle weekly working sets vs evidence-based hypertrophy bands, plus days since trained.
- get_strength_standards — rate squat / bench / deadlift / press against sex-specific population standards.
- get_training_consistency — sessions this week and month, current streak, average per week, last 8 weeks.
- get_week_review — the coach's week-in-review digest: sets, tonnage, plan adherence, PRs, focus pointers.
- assess_session — an honest verdict on a finished session, measured against your own history.
- estimate_starting_weight — a starting weight and rep range for a lift you've never logged, transferred from one you have.
- generate_plan — propose a multi-day plan with the app's own generator (goal · days · equipment), ranked against your history; read-only, persists nothing until you save it with create_routine / create_plan.
- get_cardio — logged cardio activities (runs / rides / swims), newest first.
- get_body_metrics — bodyweight, body-fat %, and custom measurements, newest first.
- search_exercises — browse the 881-movement catalog by name, muscle, or equipment.
- list_routines — your routines with each one's ordered exercises and prescription.
- list_plans — your weekly plans: which routine runs on which weekday, and the missed-day policy.
Logging & capture — write
- log_set — log one completed set onto your synced workout (idempotency-keyed, so a retry never double-logs).
- update_set — correct a logged set's weight, reps, RIR, or warmup flag.
- delete_set — delete a single mis-logged set.
- finish_workout — mark a session completed so it counts in history and stats.
- delete_workout — delete an entire session and its sets.
- log_cardio — log a run / ride / swim onto the synced timeline.
- log_body_metric — log bodyweight or a tape measurement, in the unit you enter it.
- create_exercise — create a custom exercise when a movement isn't in the catalog.
Routines & weekly plans — write
- create_routine — author a new routine; it syncs to the app and watch, ready to start.
- update_routine — rename, re-note, or fully reprogram a routine's exercise list.
- delete_routine — delete a routine (past workouts logged from it are untouched).
- create_plan — put a weekly schedule on your calendar: routines mapped onto weekdays.
- update_plan — revise the schedule, missed-day policy, or which plan is active.
- delete_plan — delete a weekly plan (the routines it scheduled stay in your list).
Profile & feedback — write
- set_profile — set sex / bodyweight for strength standards on your cloud profile.
- submit_feedback — send a bug report, feature request, or question to the MyoAmigo team, in your words.
Point your assistant at the endpoint.
You need a MyoAmigo account (sign up), then one of these. Step-by-step setup for Claude, ChatGPT, Cursor, and every other client lives in the connection guide.
claude.ai & Claude Desktop
Add a custom connector and paste the endpoint. Claude registers itself and runs the OAuth sign-in automatically.
https://api.myoamigo.com/mcp
CLIClaude Code
One command:
claude mcp add --transport http myoamigo https://api.myoamigo.com/mcp
AnyOther MCP clients
OAuth 2.1 with dynamic client registration and PKCE, or a personal API key (myo_sk_…) for headless use. See setup for ChatGPT, Cursor, VS Code, Codex, Gemini, and more.
Gated, audited, yours.
Interactive clients authenticate with OAuth 2.1 (audience-bound tokens, PKCE required); headless clients use a single personal API key you can rotate or revoke. Every tool call runs only against your account's data. Your training history is never used to train models, never sold, and never visible to another user's agent.
For developers and agents.
tools.json — the full tool catalog with input schemas · llms.txt — the endpoint, transport, and auth summary · OAuth protected-resource metadata. Prefer reading about the integration first? See AI Coach for the why, or What is an MCP server? for the background.