add_benchmark_to_session
ChatGPTAppend a benchmark test to a session. Use list_benchmarks for valid ids.
add_exercise_to_session
ChatGPTAppend an exercise from the library to a session template.
add_note
ChatGPTSave a note the climber can read back under Notes in the LEDGER app. A note is free text: it has no timers, is not logged and cannot be scheduled (use create_exercise for anything they will train). Offer to save one whenever the conversation produces something worth coming back to: a rough weekly plan, what to focus on and why, technique drills, a lesson from a benchmark result. Say what the note would contain and save it only after the climber agrees. To revise a plan saved earlier, find it with list_notes and change it with update_note instead of adding a second copy. Write the body as plain text with real line breaks and simple "-" bullets, no Markdown. Lead with the point: the climber will read it weeks later without this conversation.
archive_exercise
ChatGPTHide an exercise from the library while keeping its training history.
count_logs_per_exercise
ChatGPTHow many times each exercise has been logged. Useful for spotting neglected or over-represented work in the plan. Archived exercises are marked is_archived: true and cannot be scheduled or added to a session.
create_exercise
ChatGPTAdd an exercise to the library. The description is plain text. An exercise is something the climber trains and logs; for a written note (a plan, a focus, drills to read back later) use add_note instead. Add timers only for timed work or when the climber wants a rest timer between sets. The result's timer_summary restates the timers in words: check it against what was asked. - Repeaters, 3 sets of 6 reps, 7s on / 3s off, 3 min between sets: { "sets": [{ "count": 3, "rep_count": 6, "rep_seconds": 7, "rest_before_next_rep": 3, "rest_after_set": 180 }] } - One timed effort per set, 6 max hangs of 10s, 2 min between: { "set_based_exercise": true, "sets": [{ "count": 6, "rep_seconds": 10, "rest_after_set": 120 }] } - Efforts the climber ends by tapping, 4 lead burns, 10 min between: { "manual_trigger": true, "set_based_exercise": true, "sets": [{ "count": 4, "rest_after_set": 600 }] }
create_session
ChatGPTAdd an empty session. Add exercises and benchmarks to it afterwards.
delete_session
ChatGPTDelete a session template and its contents. Training already logged against it is not affected.
get_benchmark_logs
ChatGPTRead recorded results for one benchmark, newest first. Each benchmark records a different shape — see list_benchmarks for its unit. Read-only: results can only be written from the app.
get_user_profile
ChatGPTThe climber's profile: what they are aiming for, what kind of climber they are, whether they focus on lead or bouldering, how long they have been climbing, and their current grades. Read this first when advising what to train or how to improve, and fit the advice to their level and focus. user_type is "learning" (mostly just climbs and wants to add structure to their training) or "experienced" (already tracks their training and wants better insight into progress). Explain more for a learning climber and be more technical with an experienced one. Grades are written in the grade_system given: lead is "french" or "yds", boulder is "font" or "vscale". redpoint_grade and best_grade are the hardest they have sent in roughly the last six months. onsight_grade and flash_grade are what they send first try with about 75% certainty. Anything the climber has not filled in is null and listed in not_set. Ask them rather than assuming.
list_benchmarks
ChatGPTThe fixed catalog of benchmark tests LEDGER supports, with the unit each one records. Use these ids elsewhere.
list_day_notes
ChatGPTRead the climber's diary notes (one per day, about sleep, skin, mood, conditions or how training felt) over a date range, newest day first. Useful next to list_training_logs to see how training and how they felt line up.
list_exercises
ChatGPTList the climber's exercise library, in the order they arranged it. Each exercise's timers are also restated in words in timer_summary.
list_notes
ChatGPTList the climber's notes, in the order they arranged them under Notes in the app: the ones they wrote themselves and the ones saved with add_note. Check here before adding a note, so a plan saved earlier is updated rather than copied.
list_scheduled_items
ChatGPTRead the training calendar over a date range, grouped by day and in the order planned.
list_sessions
ChatGPTList the reusable session templates. A session is a combined set of exercises (and potentially benchmarks on occasion), with its exercises in order. name is null when the climber never named it; the app then calls it "Session N" by its position.
list_training_logs
ChatGPTRead completed training, newest first, with the per-set detail recorded at the time. Read-only: logs can only be written from the app. History includes archived exercises (is_archived: true), which cannot be scheduled or added to a session.
log_benchmark
ChatGPTRecord the result of a benchmark the climber took but didn't log, on the day they took it. A day holds one result per benchmark, and this never replaces one: use update_benchmark_log to correct it. Critical force can't be logged here, as its results come from taking the test in the app. Only record what the climber told you they did, never an estimate or a value you made up. Say what you will save and save it only after the climber agrees. Training can only be logged up to today. Work out today's date in the climber's own time zone, not UTC, and never use a day that is still to come.
log_exercise
ChatGPTRecord an exercise the climber did but didn't log, on the calendar day they did it. If it was planned that day this logs that entry, otherwise it adds it to the day. Check with list_training_logs first so it isn't logged twice. Only record what the climber told you they did, never an estimate or a value you made up. Say what you will save and save it only after the climber agrees. Training can only be logged up to today. Work out today's date in the climber's own time zone, not UTC, and never use a day that is still to come.
propose_exercises
ChatGPTShow the climber one or more exercises to review before anything is saved. Nothing is written. A client that supports it renders a card per exercise (name, category, description, timers and what will be logged); any other client gets the same as text. Takes the same fields as create_exercise, with the same rules. Add timers only for timed work or when the climber wants a rest timer between sets. The result restates the timers in words: check it against what was asked. - Repeaters, 3 sets of 6 reps, 7s on / 3s off, 3 min between sets: { "sets": [{ "count": 3, "rep_count": 6, "rep_seconds": 7, "rest_before_next_rep": 3, "rest_after_set": 180 }] } - One timed effort per set, 6 max hangs of 10s, 2 min between: { "set_based_exercise": true, "sets": [{ "count": 6, "rep_seconds": 10, "rest_after_set": 120 }] } - Efforts the climber ends by tapping, 4 lead burns, 10 min between: { "manual_trigger": true, "set_based_exercise": true, "sets": [{ "count": 4, "rest_after_set": 600 }] } Once the climber agrees, call create_exercise with the same fields for each exercise. If they want something changed, propose again with the change instead of saving.
remove_session_item
ChatGPTRemove one exercise or benchmark from a session, using the item_id from list_sessions.
reorder_sessions
ChatGPTSet the order of the sessions. Pass every session id, in the order you want them.
save_day_note
ChatGPTWrite the climber's diary note for one day: how they slept, skin, mood, conditions, how the session felt. A day has at most one diary note. If the day has none yet this creates it; if it has one, its whole body is replaced, so read it with list_day_notes first and send the full new text. This can overwrite something the climber wrote themselves. Say what the note will say and save it only after the climber agrees. Write plain text with real line breaks and simple "-" bullets, no Markdown.
schedule_benchmark
ChatGPTPut a benchmark test on a specific day in the calendar. Use list_benchmarks for valid ids.
schedule_exercise
ChatGPTPut an exercise from the library on a specific day in the calendar.
schedule_session
ChatGPTCopy every exercise and benchmark from a session onto a day, keeping their order. Archived exercises are left off the day and listed under skipped_archived.
unschedule_item
ChatGPTTake one exercise or benchmark off a day, using the schedule_id from list_scheduled_items.
update_benchmark_log
ChatGPTCorrect a saved benchmark result: its day, values or notes. Find it with get_benchmark_logs. What you leave out stays as it is; routes, when sent, replaces all of its routes, so send every route. A day holds one result per benchmark, so moving it onto a day that already has one is refused. Results can't be deleted here, and critical force can't be changed here. Only record what the climber told you they did, never an estimate or a value you made up. Say what you will save and save it only after the climber agrees. Training can only be logged up to today. Work out today's date in the climber's own time zone, not UTC, and never use a day that is still to come.
update_exercise
ChatGPTChange an existing exercise. Only the fields you pass are modified. Passing timers replaces the whole timer setup; see create_exercise for its shape.
update_exercise_log
ChatGPTCorrect a saved exercise log: its day, its sets or its notes. Find log_id with list_training_logs. What you leave out stays as it is; sets, when sent, replaces all of the log's sets, so send every set. A new date moves it on the calendar too. Logs can't be deleted here. Only record what the climber told you they did, never an estimate or a value you made up. Say what you will save and save it only after the climber agrees. Training can only be logged up to today. Work out today's date in the climber's own time zone, not UTC, and never use a day that is still to come.
update_note
ChatGPTChange the title and/or body of an existing note, found with list_notes. Only the fields you pass are changed; the body is replaced as a whole, so send the full new text. This can overwrite something the climber wrote themselves. Say what you will change and do it only after the climber agrees. Same rules for the body as add_note: plain text, real line breaks, simple "-" bullets.