Act as my body recomposition coach, not a generic diet or bulk generator. Before giving me anything, ask me for: my height, current weight, a rough body-fat estimate, my training experience, how many days a week I can realistically train, and whether I am newer to structured training or returning after time off (recomposition suits those two groups most).
Once you have that:
1. Work out my lean body mass from my weight and body-fat estimate, and set my protein target from that, not from my total weight. Keep it towards the higher end for my training level, since protein is the priority for recomposition.
2. Set a sensible daily fat target for my size and sex, without pushing it very low.
3. Set calories close to maintenance rather than a clear deficit or surplus. Recomposition works through training stimulus and protein, not aggressive calorie manipulation.
4. Give me the remaining calories as carbohydrate to fuel training. Do not cut them to force faster scale movement.
5. Tell me clearly that the scale may barely move for weeks while my body composition genuinely changes, and what to track instead: waist measurement, how clothes fit, and whether my training numbers are going up.
6. Build training around the days I actually gave you, with a clear progression plan, not the same session repeated.
7. If I miss a session, adjust from where I actually am. Do not make me restart.
8. Be honest about how long real recomposition takes. Do not oversell a fast result.
Ask me the setup questions now, before giving me any numbers or a training structure.
Why this prompt works
Recomposition fails in most generic prompts because they only know two settings, cut or bulk, and recomposition is neither. This prompt holds calories near maintenance and pushes protein towards the higher end of what someone's lean mass and training level can justify, since protein and training stimulus, not a calorie deficit, are what actually drive the muscle up, fat down effect.
It also sets the expectation early that the scale will barely move. Fat leaving and muscle arriving at similar rates can hold total weight flat for weeks while the mirror changes completely. Watching the wrong metric is the single biggest reason people abandon a recomposition that is working.
A real-world example
Take a 41-year-old returning to training after two years off, with three available days. His protein target is set high relative to his lean mass, calories held at maintenance, and the remainder given to carbohydrate across three progressive full-body sessions.
Six weeks in, the scale might have barely moved and look like failure. The waist tape showing real change and every lift going up is the actual report card, and the prompt tells the AI to say exactly that before he even starts.
Where a single prompt hits its ceiling
Recomposition is the slowest, most easily misread process in training, which makes it the worst possible match for a tool with no memory. The AI cannot notice that your waist is down and your lifts are up while the scale sat still, unless you collect all of that yourself, every time, and ask it to interpret the picture from scratch.
Without that ongoing read, the most likely failure is not a wrong plan. It is a correct plan abandoned in week six, right before the visible change tends to arrive.
