Once you connect Claude, ChatGPT, or another AI tool to FitMetrics via MCP, the most important question becomes: what do you ask it? This article is a prompt playbook — real coaching use cases you can copy-paste into your AI tool today.
All of these prompts work out of the box with a FitMetrics MCP API key. The AI tool will chain the right read and write tools automatically.
New to MCP? Start with our setup guide first.
Morning coaching routine
A 5-minute briefing before you open your inbox.
"Give me a morning briefing: which of my clients are waiting for a reply, who needs follow-up, and who had a great workout yesterday?"
"Which clients missed their daily nutrition log yesterday?"
"Summarize all AI form submissions from the last 24 hours that haven't been reviewed."
"Pull my team's activity log for the last 24 hours — who did what?"
Prepping for a client check-in
Before you message a client, have the AI assemble everything you need to know.
"I'm about to check in with Sarah. Pull her last 2 weeks: compliance, nutrition averages, workouts completed, any missed sessions, and the last 10 messages. Then draft a check-in message."
"Summarize Mike's Zoom call from last Tuesday — action items, commitments, and anything I promised to follow up on."
"What did John say in his most recent AI form submission, and how does his current compliance match what he said?"
"Read Kate's WHOOP data and nutrition for last week. Is there a recovery/intake mismatch I should flag?"
Updating a client's macros (new)
The AI can adjust a client's Trainerize macros directly. The change is pushed to Trainerize instantly and shows up in the client's app for their next meal log.
"Xenios was 400 calories over target every day this week. Drop him to 1800 calories with 200g protein."
"Set Sarah to 2200 calories, 180g protein, 250g carbs, 60g fat."
"Increase Mike's calories by 150, keep protein the same, add the difference to carbs."
"Look at the last 4 weeks of John's weigh-ins and nutrition. If he's losing too fast, bump calories up by 100. If too slow, cut by 100."
Safety: You'll see the AI's recommended change before it executes in tools that preview tool calls (like Claude Desktop). The change is logged with a full audit trail.
Scheduling macro cycles (new)
FitMetrics can schedule future macro changes and apply them automatically at a date and time you choose (in your business's timezone). Use this for diet breaks, refeed weeks, phase transitions, and planned cut/bulk bumps.
"Put Sarah on a 4-week cut starting next Monday at 7am — 1800 calories, 200g protein, 120g carbs, 60g fat. Four weeks later, schedule a refeed week at 2400 calories with 200g protein."
"Schedule a diet break for every client on a cut: two weeks from today, bump them up 300 calories. Two weeks after that, drop them back to their original target."
"Show me every pending scheduled macro change for John — I want to make sure there's nothing conflicting with the new plan."
"Cancel all scheduled macro changes for Mike — we're rethinking his plan."
"Look at Sarah's compliance over the last 4 weeks. If she's hitting above 85%, schedule a -100 calorie drop for next Monday. If below, schedule a +100 bump instead."
How it works: The AI uses schedule_macro_change to queue the change, list_scheduled_macro_changes to see what's pending, and cancel_scheduled_macro_change to remove one. FitMetrics's hourly cron applies the change at the exact local date and time and pushes the new goal to Trainerize.
Saving coach-visible notes (new)
Give the AI durable memory by asking it to save notes to the client's profile. These appear in the FitMetrics Messages > Notes panel and are shared across all message threads.
"Review Sarah's last 2 weeks of nutrition and workouts, then save a pinned note summarising what's working and what to change."
"Write a note on John's profile: 'Reported left knee discomfort on 2026-04-17 — swap barbell squats for goblet or split squats this week.'"
"After every AI form submission, save a one-paragraph summary as a note on that client so I can see context without re-reading the form."
"When I ask 'what does the coach need to know about Kate', pull her pinned notes first."
Pipeline, tasks, and follow-ups (new)
Use your connected AI to manage CRM state — moving clients through your pipeline, flagging for follow-up, and keeping a clean task list.
"Move Sarah to the Onboarded stage and create a task for me to call her next Friday."
"Read Mike's last 5 messages. If anything sounds like a concern, flag him for follow-up."
"Create a task: 'Review John's form submissions from this week', priority high, due tomorrow, assigned to me."
"Go through my to_do tasks from before today and mark any that are still relevant as in_progress; complete the rest."
"For every client whose program expires this month, flag them for follow-up AND create a renewal task."
"List every client who has been stuck in the 'Questionnaire' stage for more than 7 days and flag them for follow-up."
How it works: Uses move_client_to_stage, update_client_followup, create_task, and update_task_status. Pair with list_stages and list_tasks to read current state before changing it.
Habits (new)
FitMetrics can assign and remove Trainerize habits for your clients. Habits appear in the client's Trainerize app and are tracked daily with streak counts. Your connected AI can see them, set them up, and tear them down.
"Assign the 'Eat vegetables at every meal' habit to Sarah for 4 weeks, Monday through Friday."
"For every client on a cut, assign the 'Drink only zero calorie drinks' habit for 8 weeks."
"Show me Mike's current habits. If any have hit a 30-day streak, remove them and tell me which ones so I can celebrate with him."
"Create a custom habit for Kate: '10-minute morning walk', starting tomorrow, 6 weeks, every day."
"Audit all my clients' current habits — anyone with more than 3 active habits should drop their lowest-streak one."
How it works: get_client_habits lists current/upcoming/past habits with streak data, assign_habit_to_client adds a habit with type + name + schedule + duration, and remove_habit_from_client deletes one (validated against the client). Common habit types: eatVeggie, eatProtein, drinkOnlyZeroCalorieDrink, prepareYourOwnMeal, practiceEatingSlowly, prioritizeSelfCare, digitalDetoxOneHourBeforeBed, and customHabit for anything custom.
Bulk / cohort actions
Coaching leverage — one sentence, many clients.
"Send a congratulations message to every client who hit a workout PR this week."
"Find all clients whose program expires in the next 14 days and schedule a re-sign message for each of them for Monday at 10am."
"List every client below 70% compliance for the last 4 weeks and schedule a check-in AI template to each."
"For every client with the tag 'cutting', review their weight trend and adjust calories by no more than 100 in either direction."
Weekly macros: training vs rest, refeeds, carb cycling (new)
One step up from daily macros: your connected AI can set different calories and macros for each day of the week. Great for training vs rest day splits, refeed days, and carb cycling blocks.
"Put Sarah on a training/rest split: Mon/Wed/Fri 2500 cal 220P 280C 60F, Tue/Thu/Sat/Sun 2000 cal 220P 160C 70F."
"Set John's macros so Sunday is a refeed day at 2800 calories, and every other day stays at his current 2200."
"Carb cycle for Mike for the next 4 weeks — high carb Mon/Wed/Fri, low carb the rest, all at 200g protein."
How it works: set_client_weekly_macros takes a daily_targets object with all 7 days. Mode (full_macros / calories_protein / calories_only) is auto-detected based on what you send and must be consistent across all 7 days. The weekly plan is saved to FitMetrics and today's target is pushed to Trainerize; subsequent days apply automatically by the daily sync.
AI form submission review (new)
When an AI agent reads, summarizes, and responds to an AI form submission, have it close the loop by marking the submission reviewed — so human coaches don't double-handle.
"For each unreviewed AI form submission from this week: read the responses, write a short summary, draft a reply message to the client, and mark the submission reviewed when done."
"Un-mark Sarah's form submission from yesterday — I want to re-review it myself."
"Mark every submission older than 7 days as reviewed so my queue is clean."
How it works: list_form_submissions with reviewed: "not_reviewed" to find candidates, then mark_submission_reviewed(submission_id) to close each. Pass reviewed: false to unmark.
Managing messaging workflows
"Draft a birthday message for every client with a birthday this week, and schedule each for 9am on their birthday in their local timezone."
"Reply to Mike's last message with something encouraging and reference a specific workout he completed yesterday."
"Check my scheduled messages for this week and cancel any that look like duplicates."
AI check-in templates & automations
Meta-coaching: use your connected AI to build the AI check-in templates that automatically message your clients later.
"Create an AI check-in template called 'Weekly Nutrition Review' that uses calorie data and gives one actionable tip."
"Assign the 'Motivational Coach' AI check-in template to every client with less than 60% compliance, send every Tuesday at 9am."
"Enable the daily Trainerize sync automation at 6am in my timezone."
"Turn off the weekly Trainerize summary automation until I re-enable it."
Diagnostics & research
Use your connected AI as a read-only analyst over your whole business.
"Which coaches on my team have the highest and lowest average client compliance for the last 30 days?"
"Cross-reference WHOOP recovery with workout adherence across all my clients — is there a correlation?"
"Pull the transcripts of my last 10 Zoom calls with clients and tell me what the most common objections or friction points were."
"How many messages per week do I send on average? How does that compare to my team members?"
Tips for getting the best results
Ask the AI to chain steps. Instead of 'get Sarah's data' then 'write a note', say "pull Sarah's nutrition and workouts, then save a pinned note summarising trends". The AI will call the read and write tools in one go.
Use names the AI can look up. If you say 'Sarah', the AI will call
list_clients(search="Sarah")first. Full names are safest.Preview write actions before they run. Claude Desktop shows each tool call and asks for confirmation before executing. Review before you approve.
Use coach-scoped API keys for sensitive bulk actions. That way the AI can only touch clients assigned to a specific coach.
Create read-only API keys for analytics. When you only want to ask questions, give the AI a key with no write tools enabled.
See also
Workout programs (new)
FitMetrics now exposes the full Trainerize program-editing surface to your AI: building workouts, editing exercises, scheduling them on a client's calendar, copying phases, assigning whole programs. Below are real prompts that work end-to-end with Claude (or any MCP client).
Browsing the exercise library
"Show me 10 chest exercises that use just dumbbells."
"Find all hack squat variations in my exercise library."
"What intermediate-level back exercises are tagged as compound lifts?"
Building a program from scratch
"Build Sarah a 4-week upper/lower hypertrophy plan starting Monday. Four workouts a week. Use supersets where it makes sense and keep each workout under 60 minutes."
"Create a 6-week strength block for Mike: 3 days a week, focus on compound lifts, RPE 7-8, 5x5 on the main lifts."
"Design a 2-day full-body program for John, our beginner. Bodyweight + dumbbells only, 30 minutes per session."
Editing an existing program
"Look at Sarah's current program and swap her barbell back squats for hack squats across every workout."
"Add a 4th set to bench press in Mike's upper-body workouts."
"Change the rest time on John's deadlift to 3 minutes everywhere."
"Replace the second exercise in Kate's leg day with a leg press."
Scheduling and calendar moves
"Move John's Tuesday workout to Wednesday for the next 4 weeks."
"Schedule a deload week for Sarah starting next Monday — same workouts but drop sets by one."
"Cancel Mike's workout on the 15th, he told me he's travelling."
"Show me what's on Kate's calendar for the next 14 days."
Phases and program assignment
"Duplicate Phase 1 of Kate's program as 'Phase 2' with the same workouts but renamed, then increase every main-lift set by one."
"Assign Sarah's 8-week PPL program to John starting next Monday — addon to his existing plan."
"Add a new 2-week deload phase to Mike's program after his current phase ends."
Safety: every workout / program / calendar write requires a client id and verifies the resource belongs to that client before forwarding to Trainerize. Coach-scoped API keys are blocked from touching another coach's clients. All writes appear in the FitMetrics audit log.
Two scheduling worlds: for everyday client-calendar changes use prompts that name a specific date ("on the 15th", "next Tuesday"). The AI will use update_daily_workout / delete_daily_workout, which work on any assigned client. The "schedule_workout / move_scheduled_workout / unschedule_workout" tools work on program-template plans only — useful when you're building a reusable template program.
Food log accountability (new)
FitMetrics now exposes itemized food log data via MCP — not just daily macro totals, but the actual meals, foods, brands, amounts, and timestamps your clients log. Pair with send_message to do real-time "soft accountability" — spot off-plan meals, nudge clients gently.
Visibility — what did they eat?
"What did Sarah eat for dinner last night?"
"Show me everything Mike logged this week with timestamps."
"Which of my clients ate after 22:00 this week?"
"Anyone logged alcohol or fast food today?"
"Compare John's meal timing this week vs last week — is he eating earlier?"
Soft accountability nudges
Chain get_client_meals with send_message — the AI reads, reasons, then drafts a short on-brand message. Coach reviews + sends in one prompt.
"Sarah went 30g over carbs today — DM her something soft about swapping dinner for a lighter snack option."
"Anyone over 150% of their carb target today? Send each a low-key check-in: 'noticed your carbs ran high, all good?'"
"For every client who logged a meal after 22:00 last night, ask if they want help planning earlier dinners this week."
"Sarah didn't log lunch today and her last meal was breakfast at 7am. DM her: 'just checking in, are you eating?'"
Diagnostics + insights
"Mike says he's eating clean but stalled — pull his last 14 days of meals and tell me what stands out."
"Find clients whose protein is concentrated in one meal (>60% in a single sitting). They're probably under-eating earlier."
"Which clients hit their macros AND took a meal photo for every meal this week? They deserve a shoutout."
"Sarah eats wildly different things on weekends vs weekdays. Summarize the pattern."
Note on real-time: the AI reads on-demand when you ask. For autonomous auto-nudging (the AI fires DMs the moment a client logs a meal, no coach involvement), that's a separate "Smart Nudges" automation feature — not part of MCP itself. Today, MCP-powered accountability is reactive: you ask, the AI checks + drafts.
Workout detail + comments accountability (new)
get_client_workouts now returns the full per-exercise + per-set picture (reps, weight, time, distance — only the stats each exercise actually tracks) plus inlined comments from the workout (the client's feedback, the coach's notes, and the RPE the client logged with each comment). Same compact shape as get_client_meals — one call gets you the whole story.
Performance + progression
"Show me Mike's bench press progression over the last 4 weeks — sets x weight per session."
"Did Sarah go up in weight on her squat this week vs last week?"
"Find clients who skipped > 50% of sets on their last workout."
"What's John's heaviest set ever for deadlift? When was it?"
Pain + fatigue + comments
"Did anyone mention pain, soreness, or fatigue in their workout comments this week?"
"Pull every workout where Sarah logged an RPE of 9 or 10 — she might be over-reaching."
"Read all the comments on John's workouts this month and summarize the themes."
"Mike commented on his last workout — did I reply? Draft a quick acknowledgement if I haven't."
Engagement + adherence
"Which clients did a workout but didn't leave a comment OR an RPE?"
"List clients who had a workout scheduled yesterday but didn't log it."
"Did Sarah do Tuesday's workout on Tuesday or did she shift it? Compare scheduled date vs started_at."
"Find clients whose workouts are getting shorter over time — duration trend down 3+ weeks running."
Soft nudges from workout signals
"Sarah hit her best squat ever yesterday (180 lb x 5 — was 175 last week). Send her a hype DM."
"Mike's last 3 workouts all had RPE 9+. DM him to suggest a deload week."
"Find every client who didn't log a workout this week and send a low-pressure check-in."
Engagement at scale (new)
React to every workout, every meal log — in seconds, in bulk. Comments + reactions appear directly in the Trainerize feed your client sees.
"Read every workout from yesterday and leave a 1-line comment on each — celebrate PRs, ask about RPE 9+ ones."
"For every meal photo Sarah logged this week, drop a thumbsUp reaction."
"Mike just posted a comment on his deadlift workout — react with highFive and reply with a quick comment."
"For every client who logged a workout today, comment something specific to what they did."
Segmentation, tags, and bulk actions (new)
Tags are the segmentation primitive. Apply them, query them, then run mass DMs / form sends / habit assigns by tag.
"Tag every client on a cut with phase:cut. Then DM all of them: 'how's hunger today?'"
"List my response time analytics for the last 30 days. Who's dropping below average?"
"Send the intake form to every client tagged new:onboarding."
"Email every client tagged paid:annual a renewal-reminder note 30 days before their renewal date."
Workflow tools — meal plans, goals, message edits (new)
"Copy Mike's Mediterranean meal plan to John, Sarah, and Kate."
"Generate Sarah's weekly summary now and DM it to her."
"Set John a weight goal of 75kg by July 1st, losing 0.4kg per week."
"Add a text goal to Sarah's account: 'Sleep 8 hours every weeknight'."
"My last DM to Mike had a typo — fix it. (give me a corrected version)"
Building & editing AI Forms (new)
Your connected AI can now build, edit, publish, unpublish, and delete AI Forms directly — no separate form-builder UI required. The AI authors the form structure itself using a built-in element-type catalog (20 element types: short answer, multiple choice, rating, slider, opinion scale, ranking, grids, contact details, signature, image, html, etc.) and writes it straight to your business.
Building a form from a description
"Build me a weekly check-in form: 1-5 rating for how the week went, single-choice for did-you-train (yes/no/partial), a number question for body weight in kg, and an open-text box for anything else."
"Create an intake form for new clients: first name, last name, email, phone, date of birth, current weight, goal weight, primary goal (weight loss / muscle / recomp), and a signature field at the bottom."
"Make me a sleep + recovery form with a 1-10 sleep quality slider, opinion scale 1-7 for stress, and a multi-select for what helped sleep last night (caffeine cut-off, screens off, magnesium, walk, none)."
"Build a multi-page onboarding form: page 1 contact details, page 2 goals + experience, page 3 medical history with skip-logic, page 4 a signature."
Editing an existing form
"Open my Weekly Check-In form and add a sleep quality 1-10 slider before the open-text question."
"Add a body-fat % number field with min 5, max 50, step 0.1 to my morning check-in form."
"Change the title of my onboarding form to 'New Client Intake' and add a thank-you redirect to https://example.com/thanks after submission."
"Insert a new page with a body-photo attachment between pages 2 and 3 of my check-in form."
"Add a contact_details element with first name, last name, and email at the top of every form that's missing one."
Publish, unpublish, delete
"Publish the new check-in form so clients can submit it."
"Unpublish the old onboarding form — we're not using it anymore but I want to keep the submissions."
"Show me a delete preview for the abandoned form titled 'Test Form 2024' — how many submissions will be lost?"
"Delete the abandoned test form for real, I confirmed the cascade preview."
How it works: The AI calls list_form_element_types once to learn the catalog (every element type and its required/optional props). Then create_form with a title plus a form_content object containing pages and elements; get_form + update_form to read-modify-write existing forms; publish_form / unpublish_form to flip the published state; and delete_form with an explicit confirm: true to remove a form (a first call without confirm returns a cascade preview so the AI sees what will be lost before deleting). Coach-scoped API keys can read, build and edit forms but cannot delete — matching the in-app owner/admin-only policy.
Note: there is no separate "generate a form for me with AI" wrapper. The AI tool you connected (Claude, ChatGPT, etc.) is the generator — it designs the form using your prompt and the element catalog, then writes it directly. That keeps the design context-aware and skips a wasteful round-trip.
Forms + custom habits + smart habits (new)
"Send the weekly check-in form to every client tagged active."
"Assign a custom habit 'Bedtime by 22:30' to John, Sarah, and Mike — Mon-Fri, 4 weeks."
"Assign the Hydration smart habit to every client on a cut."
"List my smart habits and tell me which ones I haven't assigned yet."
Cardio sessions (new)
Cardio is now visible separately from strength workouts (treadmill / cycling / running / Apple Health imports).
"Did Mike do his 3 cardio sessions this week?"
"Show me Sarah's running mileage trend over the last 8 weeks."
"Find clients who skipped cardio sessions this week and DM them a low-pressure check-in."