CLI for the Hevy API.

npm:

npm install -g @caarlos0/hevy

Or run without installing via npx:

npx @caarlos0/hevy --help

Prebuilt binaries for macOS, Linux, and Windows are also available on the releases page.

Get an API key from https://hevy.com/settings?api:

export HEVY_API_KEY=...

hevy --help and hevy --version work without an API key. API commands require HEVY_API_KEY.

hevy whoami

hevy routines     list | get <id> | create --file f.json | edit <id> [--file f.json]
hevy folders      list | get <id> | create <title...>
hevy workouts     list | get <id> | create --file f.json | edit <id> [--file f.json]
hevy workouts     count | events [--since <iso>]
hevy exercises    list | get <id> | history <id> [--start <iso>] [--end <iso>]
hevy exercises    create --title T --type weight_reps --equipment dumbbell --muscle quadriceps
hevy measurements list | get <date> | create --file f.json | edit <date> [--file f.json]

Common flags: --page N, --page-size N, --json (raw API response).

Use --json for scripts and agents. Human-readable output is for terminals; JSON output is the raw API response printed to stdout.

Create and edit commands read a JSON object directly, not an API envelope:

hevy workouts create --file examples/workout.json --json
hevy routines edit r1 --file examples/routine.json --json
hevy measurements create --file examples/measurement.json --json

Use --file - for stdin:

hevy workouts get w1 --json \
  | jq '.title = "renamed"' \
  | hevy workouts edit w1 --file - --json

If --file is omitted, create commands read piped stdin. They fail instead of waiting when stdin is a TTY:

jq '.title = "copy"' examples/routine.json | hevy routines create --json

For routines and workouts, the CLI wraps the object for the API and strips server-managed fields (id, created_at, updated_at) before creating or editing. Measurements are sent as the object itself; measurement edits strip date from the PUT body because the date is part of the URL.

Errors are written to stderr and return a non-zero exit code.

exercises list paginates server-side; pipe through grep to filter:

hevy exercises list --page-size 100 | grep -i squat

edit without --file opens $VISUAL (falling back to $EDITOR, then vi). --file - reads stdin, so you can round-trip:

hevy routines get r1 --json | jq '.title="renamed"' | hevy routines edit r1 --file -

hevy workouts create --file workout.json expects the workout object directly (the CLI wraps it in { "workout": ... } for the API):

{
  "title": "Friday Leg Day 🔥",
  "description": "Quads focus",
  "start_time": "2024-08-14T12:00:00Z",
  "end_time": "2024-08-14T12:45:00Z",
  "is_private": false,
  "exercises": [
    {
      "exercise_template_id": "D04AC939",
      "superset_id": null,
      "notes": "Form on point",
      "sets": [
        { "type": "warmup", "weight_kg": 60, "reps": 8 },
        { "type": "normal", "weight_kg": 100, "reps": 10, "rpe": 8 },
        { "type": "normal", "weight_kg": 100, "reps": 10 }
      ]
    }
  ]
}

Find exercise_template_id with hevy exercises list | grep -i squat.

Set fields by exercise type:

• weight_reps: weight_kg, reps
• reps_only: reps
• duration: duration_seconds
• distance_duration: distance_meters, duration_seconds
• weight_distance_duration: weight_kg, distance_meters, duration_seconds

Set type is one of warmup | normal | failure | dropset. rpe (optional) is one of 6, 7, 7.5, 8, 8.5, 9, 9.5, 10. Group exercises into a superset by giving them the same integer superset_id.

The same shape works for edit. Routines and measurements use the same "object directly, no wrapper" convention — hevy <thing> get <id> --json shows the exact shape to send back.

MIT