apply_add_exercise
ChatGPTAdd one exercise to a workout day. Resolve masterId with search_exercise_library. Defaults to the athlete’s history-derived starting weight. A blacklisted exercise or unavailable equipment returns a warning to raise before approval. Save only this agreed change after the athlete approves its preview. Re-fetch and validate the program and check the preview's concurrency preconditions before writing. Keeps the current-program selection unchanged. Day and exercise indices are zero-based; resolve them from get_program_summary.
apply_add_rest_day
ChatGPTInsert one rest day after a specified day, or at the end when omitted. Use afterDayIndex -1 to insert at the beginning. Save only this agreed change after the athlete approves its preview. Re-fetch and validate the program and check the preview's concurrency preconditions before writing. Keeps the current-program selection unchanged. Day and exercise indices are zero-based; resolve them from get_program_summary.
apply_add_workout_day
ChatGPTInsert one workout day with the specified exercises, after a specified day or at the end when omitted. Use afterDayIndex -1 for the beginning. Resolve masterId values with search_exercise_library. Starting weights come from the athlete’s history. Save only this agreed change after the athlete approves its preview. Re-fetch and validate the program and check the preview's concurrency preconditions before writing. Keeps the current-program selection unchanged. Day and exercise indices are zero-based; resolve them from get_program_summary.
apply_compose_program
ChatGPTBuild a program from a week you design yourself, rather than from one of Push's structured templates. This is the alternative to preview_generate_program/apply_generate_program: use it when the athlete wants a program built around their own history, library and preferences instead of a template fill. Ask the athlete which they want before building - Push's own generator is proven and enforces its volume, frequency and fatigue-ordering rules while it builds, whereas composing gives you full control and only warns afterwards. Read search_exercise_library, get_training_setup, get_muscle_volume_report and get_push_science first, then give every day in order with its exercises most to least fatiguing. Starting weights come from the athlete's own history unless you set one. The result is checked against Push's principles - weekly volume per muscle, sets per muscle per session, frequency, heavy compounds per day, exercise order and session length - and those come back as warnings to raise with the athlete, not as refusals. This one saves it. Preview first unless the athlete has already agreed to the week you are about to save.
apply_delete_exercise
ChatGPTRemove one exercise prescription from a workout day; completed workout history is unchanged. Identify it with exactly one of exerciseId or exerciseIndex. Save only this agreed change after the athlete approves its preview. Re-fetch and validate the program and check the preview's concurrency preconditions before writing. Keeps the current-program selection unchanged. Day and exercise indices are zero-based; resolve them from get_program_summary.
apply_delete_rest_day
ChatGPTRemove one existing rest day from a saved Push program. Save only this agreed change after the athlete approves its preview. Re-fetch and validate the program and check the preview's concurrency preconditions before writing. Keeps the current-program selection unchanged. Day and exercise indices are zero-based; resolve them from get_program_summary.
apply_delete_workout_day
ChatGPTRemove one workout day and its exercise prescriptions from a saved Push program; completed workout history is unchanged. Save only this agreed change after the athlete approves its preview. Re-fetch and validate the program and check the preview's concurrency preconditions before writing. Keeps the current-program selection unchanged. Day and exercise indices are zero-based; resolve them from get_program_summary.
apply_generate_program
ChatGPTGenerate, validate, and save a Push program with the app's structured-template generator. The alternative, a week you design yourself, is apply_compose_program; the athlete chooses which. Pass the seed returned by preview_generate_program to save the same exercise selection that was previewed; starting weights are re-read from their history, so they reflect their newest session. Revalidates before writing user data. Settings not given default to the user's saved setup (see get_training_setup), and the result reports where each came from. Like the app, generation never picks the user's custom exercises or ones they have blacklisted; to include a custom exercise, add it afterwards with preview_replace_exercise or preview_add_exercise.
apply_move_day
ChatGPTMove one workout or rest day to a new position in a saved Push program. Save only this agreed change after the athlete approves its preview. Re-fetch and validate the program and check the preview's concurrency preconditions before writing. Keeps the current-program selection unchanged. Day and exercise indices are zero-based; resolve them from get_program_summary.
apply_move_exercise
ChatGPTMove one exercise to a specified position in the same or another workout day. Identify it with exactly one of exerciseId or exerciseIndex. Save only this agreed change after the athlete approves its preview. Re-fetch and validate the program and check the preview's concurrency preconditions before writing. Keeps the current-program selection unchanged. Day and exercise indices are zero-based; resolve them from get_program_summary.
apply_rename_day
ChatGPTRename one workout day in a saved Push program. Save only this agreed change after the athlete approves its preview. Re-fetch and validate the program and check the preview's concurrency preconditions before writing. Keeps the current-program selection unchanged. Day and exercise indices are zero-based; resolve them from get_program_summary.
apply_rename_program
ChatGPTRename one saved Push program. Save only this agreed change after the athlete approves its preview. Re-fetch and validate the program and check the preview's concurrency preconditions before writing. Keeps the current-program selection unchanged. Day and exercise indices are zero-based; resolve them from get_program_summary.
apply_replace_exercise
ChatGPTReplace one exercise in a workout day. Identify the original with exactly one of exerciseId or exerciseIndex and resolve the replacement masterId with search_exercise_library. Starting weights come from history; blacklist and equipment warnings require agreement. Save only this agreed change after the athlete approves its preview. Re-fetch and validate the program and check the preview's concurrency preconditions before writing. Keeps the current-program selection unchanged. Day and exercise indices are zero-based; resolve them from get_program_summary.
apply_update_exercise
ChatGPTUpdate the sets, reps, load, weight unit, rest time or progression setting of one exercise. Identify it with exactly one of exerciseId or exerciseIndex. Prescribe a weight only when the athlete requested it; otherwise leave history-based progression in control. Unit changes convert existing loads. Save only this agreed change after the athlete approves its preview. Re-fetch and validate the program and check the preview's concurrency preconditions before writing. Keeps the current-program selection unchanged. Day and exercise indices are zero-based; resolve them from get_program_summary.
duplicate_program
ChatGPTSave a copy of one of the athlete's programs under a new name, exactly as the app's "Duplicate program" does. The copy is left alongside the original rather than made current, so it is the safe way to try changes: copy the program, preview and apply the specific edit tool against the copy (passing its programId), and switch to it only when the athlete is happy. Defaults to copying the current program.
get_consistency_report
ChatGPTAnalyse Push workout consistency, current streak, longest streak, gaps, and recent training dates.
get_current_training_context
ChatGPTReturn an AI-friendly snapshot of the user, current program, recent training, last workout, muscle volume, and recovery readiness.
get_exercise_history
ChatGPTReturn detailed performed-set history for an exercise, grouped by session, including top sets and a next-target baseline.
get_exercise_progress
ChatGPTShow session-by-session progression for an exercise using completed sets and estimated 1RM.
get_muscle_volume_report
ChatGPTAggregate recent completed sets by target muscle, including direct sets, optional synergist sets, last trained dates, and top contributing exercises.
get_program_summary
ChatGPTSummarise a Push program without mutating it: workout days, exercises, target muscles, and weekly set targets. Defaults to the current program; pass a programId from list_programs to read any other one.
get_push_science
ChatGPTRead Push's own science articles: the evidence and reasoning behind how Push sets volume, frequency, exercise order, rep ranges, progression, and plateau handling. Call this before justifying a program change, or when the user asks why Push trains a certain way. Omit articleId to list what is available.
get_recovery_readiness
ChatGPTEstimate which muscles are likely ready versus worth avoiding or reducing based on recent direct set volume and last trained date.
get_training_calendar
ChatGPTReturn workout history as calendar days with workout names, exercises, target muscles, completed sets, and volume.
get_training_setup
ChatGPTThe user's saved training setup and exercise preferences: the equipment, kit tags, days, split, duration and focus their current program was generated from (with where each value came from), their blacklisted ("Never select") exercises, exercises they have re-prioritised, their custom exercises, and the equipment and kit-tag vocabulary for turning a description of their gym into generation inputs.
get_workout_summary
ChatGPTSummarise Push workouts, sets, reps, volume, PBs, consistency, and most trained exercises.
list_personal_bests
ChatGPTList PB sets and strongest estimated 1RM sets from Push workout history.
list_programs
ChatGPTEvery program on this account, newest first, with the id each program tool takes: which one they are training now, how many workouts a week it has, when it was last changed, and how it was made. Call this before reading, editing, copying or deleting anything other than the current program.
preview_add_exercise
ChatGPTAdd one exercise to a workout day. Resolve masterId with search_exercise_library. Defaults to the athlete’s history-derived starting weight. A blacklisted exercise or unavailable equipment returns a warning to raise before approval. Preview only: validates the proposed program without saving. Return the proposal and concurrency preconditions for approval. Keeps the current-program selection unchanged. Day and exercise indices are zero-based; resolve them from get_program_summary.
preview_add_rest_day
ChatGPTInsert one rest day after a specified day, or at the end when omitted. Use afterDayIndex -1 to insert at the beginning. Preview only: validates the proposed program without saving. Return the proposal and concurrency preconditions for approval. Keeps the current-program selection unchanged. Day and exercise indices are zero-based; resolve them from get_program_summary.
preview_add_workout_day
ChatGPTInsert one workout day with the specified exercises, after a specified day or at the end when omitted. Use afterDayIndex -1 for the beginning. Resolve masterId values with search_exercise_library. Starting weights come from the athlete’s history. Preview only: validates the proposed program without saving. Return the proposal and concurrency preconditions for approval. Keeps the current-program selection unchanged. Day and exercise indices are zero-based; resolve them from get_program_summary.
preview_compose_program
ChatGPTBuild a program from a week you design yourself, rather than from one of Push's structured templates. This is the alternative to preview_generate_program/apply_generate_program: use it when the athlete wants a program built around their own history, library and preferences instead of a template fill. Ask the athlete which they want before building - Push's own generator is proven and enforces its volume, frequency and fatigue-ordering rules while it builds, whereas composing gives you full control and only warns afterwards. Read search_exercise_library, get_training_setup, get_muscle_volume_report and get_push_science first, then give every day in order with its exercises most to least fatiguing. Starting weights come from the athlete's own history unless you set one. The result is checked against Push's principles - weekly volume per muscle, sets per muscle per session, frequency, heavy compounds per day, exercise order and session length - and those come back as warnings to raise with the athlete, not as refusals. This one validates and reviews without writing anything, and returns the seed that reproduces it - pass that seed to apply_compose_program to save exactly what was previewed.
preview_delete_exercise
ChatGPTRemove one exercise prescription from a workout day; completed workout history is unchanged. Identify it with exactly one of exerciseId or exerciseIndex. Preview only: validates the proposed program without saving. Return the proposal and concurrency preconditions for approval. Keeps the current-program selection unchanged. Day and exercise indices are zero-based; resolve them from get_program_summary.
preview_delete_rest_day
ChatGPTRemove one existing rest day from a saved Push program. Preview only: validates the proposed program without saving. Return the proposal and concurrency preconditions for approval. Keeps the current-program selection unchanged. Day and exercise indices are zero-based; resolve them from get_program_summary.
preview_delete_workout_day
ChatGPTRemove one workout day and its exercise prescriptions from a saved Push program; completed workout history is unchanged. Preview only: validates the proposed program without saving. Return the proposal and concurrency preconditions for approval. Keeps the current-program selection unchanged. Day and exercise indices are zero-based; resolve them from get_program_summary.
preview_generate_program
ChatGPTPreview a Push program generated by the same structured-template generator the app uses. Ask the athlete first whether they want this or a program built around their own history (preview_compose_program); this one is the proven, balanced build. This validates the result but does not write user data. Returns the seed it used: pass that seed to apply_generate_program to save this exact program. Settings not given default to the user's saved setup (see get_training_setup), and the result reports where each came from. Like the app, generation never picks the user's custom exercises or ones they have blacklisted; to include a custom exercise, add it afterwards with preview_replace_exercise or preview_add_exercise.
preview_move_day
ChatGPTMove one workout or rest day to a new position in a saved Push program. Preview only: validates the proposed program without saving. Return the proposal and concurrency preconditions for approval. Keeps the current-program selection unchanged. Day and exercise indices are zero-based; resolve them from get_program_summary.
preview_move_exercise
ChatGPTMove one exercise to a specified position in the same or another workout day. Identify it with exactly one of exerciseId or exerciseIndex. Preview only: validates the proposed program without saving. Return the proposal and concurrency preconditions for approval. Keeps the current-program selection unchanged. Day and exercise indices are zero-based; resolve them from get_program_summary.
preview_program_constraints
ChatGPTInterpret scheduling or muscle constraints against the current program and return a non-writing preview for future program generation.
preview_rename_day
ChatGPTRename one workout day in a saved Push program. Preview only: validates the proposed program without saving. Return the proposal and concurrency preconditions for approval. Keeps the current-program selection unchanged. Day and exercise indices are zero-based; resolve them from get_program_summary.
preview_rename_program
ChatGPTRename one saved Push program. Preview only: validates the proposed program without saving. Return the proposal and concurrency preconditions for approval. Keeps the current-program selection unchanged. Day and exercise indices are zero-based; resolve them from get_program_summary.
preview_replace_exercise
ChatGPTReplace one exercise in a workout day. Identify the original with exactly one of exerciseId or exerciseIndex and resolve the replacement masterId with search_exercise_library. Starting weights come from history; blacklist and equipment warnings require agreement. Preview only: validates the proposed program without saving. Return the proposal and concurrency preconditions for approval. Keeps the current-program selection unchanged. Day and exercise indices are zero-based; resolve them from get_program_summary.
preview_update_exercise
ChatGPTUpdate the sets, reps, load, weight unit, rest time or progression setting of one exercise. Identify it with exactly one of exerciseId or exerciseIndex. Prescribe a weight only when the athlete requested it; otherwise leave history-based progression in control. Unit changes convert existing loads. Preview only: validates the proposed program without saving. Return the proposal and concurrency preconditions for approval. Keeps the current-program selection unchanged. Day and exercise indices are zero-based; resolve them from get_program_summary.
search_exercise_library
ChatGPTSearch every exercise available to the user (the Push library plus their custom exercises), not just ones they have performed. Filter by muscle, movement pattern, compound/isolation, equipment, or name. By default results fit the user's saved equipment and leave out exercises Push never selects (the user's blacklist, and the ones the library itself excludes from generation). Each result has its masterId (for the program edit tools), muscles, movement patterns, priority as the app labels it, whether it is custom, whether and by whom it is blacklisted ('user' or 'library'), whether it is already in the current program, and the user's history with it (sessions, last performed). Use sort 'least-recent' or neverPerformed to find fresh options for a neglected muscle.
search_exercises_performed
ChatGPTSearch exercises the user has actually performed, with session counts, last performed date, equipment, and target muscles.
set_current_program
ChatGPTPut the athlete on one of their existing programmes. Use this to switch programmes, rather than re-saving a programme with makeCurrent - that logs a change they did not make and bumps the programme's last-edited time. Find the id with list_programs.