)

Loadstar

One portable JSON file for a whole training block: the plan written by your coach (human or AI, and the log your app appends. It travels both ways.

)

πŸ“ Plan is a contract

Goals with baselines and a test session, nutrition targets, phases (accumulation β†’ taper β†’ test), strength and cardio sessions () everything a coach decides, in one reviewable artifact.

πŸ” Results travel back

The plan section stays immutable; the app appends an append-only log. Hand the same file back to the coach for review and the next block. No accounts, no platform lock-in.

πŸ€– Bring your own coach

Paste the canonical prompt into Claude, ChatGPT or hand it to your trainer: the output is a valid block file. Validate it here, load it, train.

The lifecycle

coach (AI/human) ──plan──▢ block.gymsync.json ──load──▢ app (train, log appends)
      β–²                                                        β”‚
      └────────────────── review / next block β—€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”˜

Sixty-second tour

{
  "gymsyncBlock": 2,
  "meta":      { "id": "block-1", "label": "Block 1", "start": "2026-08-03",
                 "weeks": 6, "units": "kg" },
  "goals":     [ { "id": "bench-100", "kind": "lift1RM", "exerciseId": "bench-press",
                   "target": { "kg": 100, "reps": 1 }, "testSessionId": "W6D1" } ],
  "nutrition": { "kcal": 2500, "proteinG": 180, "trainingDayCarbsDelta": 40 },
  "exercises": [ { "id": "bench-press", "name": "Bench Press",
                   "muscles": ["chest", "triceps"] } ],
  "sessions":  [ { "id": "W1D1", "week": 1, "type": "strength",
                   "work": [ { "exerciseId": "bench-press", "sets": 4, "reps": "5",
                               "load": { "kg": 80 }, "main": true } ] },
                 { "id": "W1R1", "week": 1, "type": "run",
                   "cardio": { "kind": "zone2", "durationMin": 35, "hr": [127, 145] } } ],
  "log":       { "sessions": [], "measurements": [] }
}

Read the spec Schema & examples

Why not just a workout tracker?

Trackers log what you did. Adaptive apps decide for you and keep the plan hidden inside. This format makes the plan itself a first-class, portable document () so any coach (or any AI) can write it, any app can execute it, and the comparison of planned vs performed is always possible.