Use your own AI

I train with a workout tracker called Badger. Below (or attached) is a CSV export of every set I have logged, one row per set, with these columns:

date, workout_id, exercise_name, category, set_number, is_warmup, weight, weight_unit, reps, duration_seconds, distance, rpe, rir, comment, is_complete, is_skipped, heart_rate, weight_per_side, is_unilateral, reps_left, reps_right, duration_left_seconds, duration_right_seconds

How to read it:
- Only rows where is_complete is true were actually performed. A workout started from a routine creates all of its sets up front, so a session I began and abandoned still has rows here, filled in with what I planned to lift. Count only completed sets as training.
- is_skipped marks a set I deliberately passed on. It is never training either.
- Weights are stored in the unit named in weight_unit and are never converted, so convert before comparing kg and lbs rows.
- Rows where is_warmup is true are warmups. Leave them out of working volume.
- Empty cells mean the exercise does not use that field: a plank has duration_seconds and no reps.
- rpe is 1-10, rir is reps in reserve. Both are optional and often blank.
- heart_rate is my heart rate at the end of that set, in beats per minute, read from Apple Health or Health Connect. It is optional and usually blank: it only exists for sets logged while I was wearing a watch, with that feature switched on. Never treat a blank as zero.
- weight_per_side says what the weight column MEANS for that row: true means the number is what was loaded on one side (iso-lateral machines), false means it is the total, and blank means the row predates the setting and is unknown. The number itself is never doubled or converted, so treat a true row as roughly half the total load when you compare it with a total row, and say so when you do.
- is_unilateral is true on a set where I logged my left and right separately (one-arm or one-leg work). On those rows reps and duration_seconds hold my better arm, and reps_left, reps_right, duration_left_seconds and duration_right_seconds hold each arm. Those four columns are blank on every other row. Weight is one number for both arms, so a set's volume is weight x (reps_left + reps_right). A blank is_unilateral means an ordinary set. Never compare one-arm rows with both-arms rows of the same exercise as if they were the same lift, and point out a gap between my arms when there is one.

Please:
1. Summarize how I have actually been training: sessions per week, which lifts I do most, how volume splits across categories, and how consistent I am.
2. Show the trend on my main lifts: best working set and estimated 1RM over time, and whether each is climbing, flat, or slipping.
3. Point out imbalances, neglected movement patterns, and anything that looks like a plateau. Say what evidence you used for each.
4. Tell me what you would change, in priority order, and where you are unsure.

Ask me about my goals, my schedule, and any injuries before assuming them.

Below is a training program as a .badger file: the plain JSON format used by the Badger workout tracker. It is documented at https://badger.fit/badger-format/

Read it and tell me:
1. What this program actually is: how many sessions a week, what it prioritizes, and who it suits.
2. Where it is thin or unbalanced. Which muscle groups and movement patterns get little or no work?
3. How hard it is to recover from, and whether the volume is sane for someone training this often.
4. What you would change for my situation, and why. If I paste my gym's .badger file underneath, also say whether the program fits the equipment I actually have.

My situation: [training age, days a week I can train, goal, equipment, anything I need to work around].

Do not rewrite the file yet. Tell me what you would change first, then I will ask.

Write me a training program as a .badger file for the Badger workout tracker, so I can import it directly.

The format is documented at https://badger.fit/badger-format/ and there is a small worked example at https://badger.fit/programs/example-routine.badger

Follow the format exactly:
- Top level: "badger": "share/v1", "type": "routine", plus app_version, created_at, and payload.
- payload has routine (an object with at least "name"), days, exercises, and machines (machines can be an empty array).
- Each day has "name" and an "exercises" array of slots. Optionally add "scheduled_days": weekday numbers, 1 = Monday to 7 = Sunday.
- Each slot's "exercise_name" and "exercise_type" must exactly equal a "name" and "type" in payload.exercises. The slot's planned sets go in its "predefined_sets" array.
- Each entry in payload.exercises has "name", "type" and "category_name". type is one of: weight_reps, reps_only, time_only, distance_only, weight_only, weight_time, weight_distance, reps_time, reps_distance, distance_time.
- A set uses the fields its type needs: "reps" (and "weight" when you have one) for weight_reps, "duration_seconds" for time_only, "distance" in kilometers for distance types. Mark warmups with "is_warmup": true.
- Use ordinary exercise names ("Barbell Back Squat", not "BB squat"), so they match exercises I already have instead of creating duplicates.
- For a rep range, set reps to the bottom of the range and reps_max to the top. For an exact target, use reps alone.
- For an open-topped ("AMRAP") set, set reps to the minimum and "is_amrap": true, and leave reps_max out. That reads as "8+": at least 8 reps, then as many as possible. A set to failure is reps 1 with is_amrap true. Never set both is_amrap and reps_max on the same set.
- Leave weight out. I fill in my own weights.
- For a percentage-based set, write "percent_of_reference" (85.0 means 85%) and still leave weight out.
- Never put reps_max or is_amrap on a warmup set.
- Order matters: days, exercises, and sets import in the order they appear in their arrays.
- To pair exercises as a superset, give both slots the same "superset_key". To pre-assign swap options, use "alternatives": an array of objects like { "exercise_name": "...", "exercise_type": "..." }, each also defined in payload.exercises. For a tempo prescription, put it in "tempo" on the slot, like "3-1-2-0".
- If I paste a gym .badger file below, plan around its equipment: only pick machine exercises its machines support (barbell, dumbbell, and bodyweight movements are always available). For an exercise done on one of those machines, set default_machine_name to that machine's name exactly as written in my file, and copy the machine's entry into payload.machines verbatim. Never invent or rename a machine.
- A gym file may also list "bars" and "accessories". Read them for what the gym has, but do not put them in the routine: a routine slot carries no default bar and no accessories, so mention band or chain work in your explanation underneath instead of trying to encode it.
- Progression can live in the file. If the program changes week to week (like 5/3/1), write the day once with "cycle_length" and give each set a 0-based "week_index"; a set with no week_index applies every week. If a slot progresses by a fixed rule, you may write it as a "progression" object on the slot, following the Progression and Clause sections of the format page exactly. It needs "cadence", "reference_kind" and a "clauses" array in evaluation order that ends with an "always" clause; each clause needs "trigger", "action" and "target", plus "amount" (add_absolute, set_from_remembered), "percent" (add_percent) or "reps_min" (driver_reps_at_least, driver_reps_between) where the format page says so. Mark the set a driver trigger reads with "is_progression_driver": true. A rule that breaks those rules is dropped on import while the rest of the file still imports, so when unsure, leave the rule out and describe it underneath instead.

The smallest valid file looks like this:
{
  "badger": "share/v1",
  "app_version": "1.1.12",
  "created_at": "2026-09-23T00:00:00Z",
  "type": "routine",
  "payload": {
    "routine": { "name": "My Program" },
    "days": [
      {
        "name": "Day A",
        "exercises": [
          {
            "exercise_name": "Barbell Back Squat",
            "exercise_type": "weight_reps",
            "predefined_sets": [
              { "is_warmup": false, "reps": 5 }
            ]
          }
        ]
      }
    ],
    "exercises": [
      { "name": "Barbell Back Squat", "type": "weight_reps", "category_name": "Legs" }
    ],
    "machines": []
  }
}

What I want:
- Days a week: [4]
- Experience: [about two years of consistent lifting]
- Goal: [get stronger on the main barbell lifts while keeping some size work]
- Equipment: [full gym: rack, barbell, dumbbells, cables, common machines]
- Session length: [about an hour]
- Avoid: [nothing]

Give me the complete JSON in one code block, then explain the program in a few sentences underneath: what progresses, how, and when I should change something.