Agent-Skills.md

Agent Skills: Fitness Tracker

Log and analyze workouts. Use when the user wants to (1) record a workout at the gym via conversation, (2) parse workout markdown files to CSV, (3) query exercise history with DuckDB, or (4) track fitness progress. Triggers on "log workout", "record workout", "track workout", "add exercises", "what did I lift", etc.

UncategorizedID: mberg/claude-skills/fitness-tracker

Repository

mberg/claude-skills
mbergLicense: CC0-1.0

Install this agent skill to your local

pnpm dlx add-skill https://github.com/mberg/claude-skills/tree/HEAD/skills/fitness-tracker

Skill Files

Browse the full folder contents for fitness-tracker.

Download Skill

Loading file tree…

skills/fitness-tracker/SKILL.md

Skill Metadata

Name
fitness-tracker
Description
Log and analyze workouts. Use when the user wants to (1) record a workout at the gym via conversation, (2) parse workout markdown files to CSV, (3) query exercise history with DuckDB, or (4) track fitness progress. Triggers on "log workout", "record workout", "track workout", "add exercises", "what did I lift", etc.

Fitness Tracker

Log workouts via conversation and save to github and analyze workout data with DuckDB.

Chat Mode - Logging Workouts

Use this conversational flow when the user wants to record a workout:

Step 1: Date

Ask for the workout date or default to today. Format: YYYY-MM-DD

Step 2: Muscle Group

Ask what they're working. Categories from reference/exercises.md:

  • Chest, Back, Shoulders, Legs
  • Arms - Biceps, Arms - Triceps
  • Core, Compound, Cardio

Step 3: Exercise Selection

Show exercises from that category. Let user pick or type their own.

Step 4: Sets Entry

User enters sets naturally:

  • "135 for 10, 10" → - 135 - 10, 10
  • "185 - 8, 8, 8" → - 185 - 8, 8, 8
  • "10, 10, 10" (bodyweight) → - 10, 10, 10
  • "30 minutes" (cardio) → - 30m
  • Add "warmup" or other notes as // comment

Step 5: More Exercises?

Ask if they want to add another exercise. Repeat steps 2-4 as needed.

Step 6: Save

Save the workout to GitHub (or Obsidian as fallback):

  • File: YYYY-MM-DD.md (one file per day, e.g., 2026-01-02.md)
  • If file exists, append new exercises (multiple workouts per day are supported)
  • Push to GitHub using scripts/github_sync.py

Example Conversation

User: Log my workout
Claude: What date? (default: today, 2026-01-02)
User: Today
Claude: What are you working? (Chest, Back, Shoulders, Legs, Arms, Core, Cardio)
User: Chest
Claude: Chest exercises: Bench Press, Incline Bench Press, Dumbbell Press, Cable Flyes, Push-ups, Dips...
User: Bench
Claude: Enter your sets (e.g., "135 - 10, 10" or "185 for 8, 8, 8"):
User: 135 for 10, 10 warmup, then 185 - 8, 8, 8
Claude: Got it:
  Bench Press
  - 135 - 10, 10 // warmup
  - 185 - 8, 8, 8

  Another exercise?
User: Cable flyes, 30 - 12, 12, 12
Claude: Added. Another exercise?
User: Done
Claude: Saved to 2026-01-02.md

Workout File Format

Files are named YYYY-MM-DD.md (one file per day). Multiple workouts in a day append to the same file.

Example: 2026-01-02.md

Bench
- 135 - 10, 10 // warmup
- 185 - 8, 8, 8

Situps
- 10, 10, 10

Treadmill
- 30m

Format Rules

  • Filename: YYYY-MM-DD.md (date is in the filename)
  • Exercise name: Plain text on its own line
  • Weighted sets: - weight - reps, reps, reps (e.g., - 185 - 8, 8, 8)
  • Bodyweight sets: - reps, reps, reps (e.g., - 10, 10, 10)
  • Duration: - 30m or - 1h30m
  • Notes: // comment at end of line

Configuration

Edit scripts/config.json to configure GitHub (default) or Obsidian storage:

{
  "github_repo": "username/repo-name",
  "github_token": "github_pat_xxxxx",
  "github_branch": "main",
  "github_workout_dir": "workouts",
  "obsidian_workout_dir": "/path/to/obsidian/vault/Fitness"
}

GitHub Configuration (Default)

  • github_repo: Your repository in username/repo-name format
  • github_token: Personal Access Token with repo scope. Create at: https://github.com/settings/tokens
  • github_branch: Target branch (default: main)
  • github_workout_dir: Directory in repo for workout files (default: workouts)

Obsidian Configuration (Fallback)

  • obsidian_workout_dir: Local path to Obsidian vault directory (used if GitHub not configured)

Writing Workout Files

When saving workouts in chat mode:

GitHub Workflow

Important: Must cd into the scripts directory so github_sync.py can find config.json.

Step 1: Fetch existing file from GitHub (if any)

cd /mnt/skills/user/fitness-tracker/scripts
uv run github_sync.py --fetch workouts/2026-01-02.md > /tmp/2026-01-02.md

If the file doesn't exist yet, create it empty.

Step 2: Append exercises to the local file

cat >> /tmp/2026-01-02.md << 'EOF'

Pull-ups
- 10, 10, 10
EOF

Step 3: Push back to GitHub

cd /mnt/skills/user/fitness-tracker/scripts
uv run github_sync.py \
  --file /tmp/2026-01-02.md \
  --dest workouts/2026-01-02.md \
  -m "Add workout for 2026-01-02"

Step 4: Update CSV

Fetch existing CSV, parse the new workout, and update:

cd /mnt/skills/user/fitness-tracker/scripts
uv run github_sync.py --fetch workouts/workouts.csv > /tmp/workouts.csv
uv run parse_workout.py /tmp/2026-01-02.md -o /tmp/workouts.csv --append
uv run github_sync.py \
  --file /tmp/workouts.csv \
  --dest workouts/workouts.csv \
  -m "Update workout data"

The --append flag replaces rows for the parsed date(s) while keeping other dates intact.

Obsidian (Fallback)

If GitHub is not configured (github_repo is empty):

  1. Use obsidian_workout_dir from config
  2. Target file: {obsidian_workout_dir}/YYYY-MM-DD.md
  3. Use Obsidian MCP tools or write directly to filesystem

Parse Mode - Analyzing Workouts

# Parse all daily workout files in a directory
uv run scripts/parse_workout.py /path/to/workouts/ -o workouts.csv

# Parse a single day and update existing CSV (replaces that date's rows)
uv run scripts/parse_workout.py /tmp/2026-01-02.md -o workouts.csv --append

# Parse and run a query
uv run scripts/parse_workout.py /path/to/workouts/ --query "SELECT * FROM workouts"

The parser matches files named YYYY-MM-DD.md and extracts the date from the filename.

Exercise Validation

The parser checks exercise names against reference/exercises.md and suggests corrections for typos:

Unrecognized exercises:
  'Bech Press' - did you mean 'Bench Press'?
  'Squatts' - did you mean 'Squats'?

Add new exercises to reference/exercises.md to expand the known list.

Output CSV Schema

| Column | Type | Description | |--------|------|-------------| | date | DATE | Workout date (YYYY-MM-DD) | | exercise | VARCHAR | Exercise name | | set_num | INTEGER | Set number within exercise | | weight | DECIMAL | Weight used (empty for bodyweight) | | reps | INTEGER | Rep count (empty for duration exercises) | | duration_min | INTEGER | Duration in minutes (for timed exercises) | | notes | VARCHAR | Comments from // note |

Example Queries

-- Total volume per exercise
SELECT exercise, SUM(weight * reps) as volume
FROM workouts GROUP BY exercise ORDER BY volume DESC;

-- Bench press max weight over time
SELECT date, MAX(weight) as max_weight
FROM workouts WHERE exercise = 'Bench' GROUP BY date;

-- Weekly workout frequency
SELECT DATE_TRUNC('week', date) as week, COUNT(DISTINCT date) as days
FROM workouts GROUP BY 1 ORDER BY 1;

-- Exercises by total sets
SELECT exercise, COUNT(*) as total_sets
FROM workouts GROUP BY exercise ORDER BY total_sets DESC;