Analyze Fitness Data and Generate Insights

Fitness Analyzer is a skill, written in Chinese and maintained by WellAlly Tech, that analyzes exercise data, identifies exercise patterns, assesses fitness progress and produces personalized training suggestions. It can also correlate exercise data with chronic-condition data. Its tool access is limited to Read, Grep, Glob and Write.

It has five functions. Trend analysis looks at exercise volume (duration, distance, calories), weekly frequency, the share of low, medium and high intensity, and changes in preferred exercise type, and reports the trend direction (improving, stable or declining), the size of the change, its significance and suggestions. Progress tracking covers running (pace, distance, heart rate), strength training (weight, volume, RPE), endurance and flexibility, reporting start versus current values, percentage improvement and milestones.

Habit analysis looks at preferred exercise times of day, weekly frequency patterns, exercise type preferences, rest day distribution and a consistency score from 0 to 100. Correlation analysis relates exercise to weight, blood pressure, blood glucose, and mood and sleep, returning a correlation coefficient from -1 to 1, its strength, significance, causal inference and practical advice. Personalized suggestions cover frequency, intensity, exercise type, timing and recovery, based on WHO, ACSM and AHA guidelines, the user's history, health status and goals.

Use it to analyze exercise records, training intensity, exercise habits or fitness progress, for trend and improvement advice across running, strength, endurance or flexibility, and when exercise data needs to be correlated with other health modules.

The skill sets a medical safety boundary. Its analysis is for health reference only and is not medical advice. It can do exercise statistics, trend identification, correlation calculation and general suggestions. It does not diagnose disease, assess exercise risk, design specific exercise prescriptions or diagnose and treat sports injuries. It also flags warning signs such as exercise heart rate above 95% of maximum, resting heart rate above 100 bpm, systolic pressure of 180 mmHg or more or diastolic of 110 or more, seven consecutive days of high-intensity exercise, RPE above 17, and weight loss above 1 kg per week. Advice comes in three levels: general (guideline-based), reference (based on user data) and medical (needs a doctor's confirmation).

Inputs are local data files: exercise logs at data/fitness-logs/YYYY-MM/YYYY-MM-DD.json, the user profile at data/fitness-tracker.json, and linked health data in data/hypertension-tracker.json, data/diabetes-tracker.json and data/profile.json. A data quality check verifies completeness, plausible ranges, timestamp consistency and duplicates.

Outputs are Markdown reports: a trend report with volume, frequency, intensity and type distribution; a correlation report (for example exercise time versus systolic pressure); and a running progress report with pace, distance, heart rate and milestones. The algorithms are linear regression (y = a + bx) for trends, the Pearson correlation coefficient (weak below 0.3, moderate to 0.7, strong from 0.7), pace as duration divided by distance, and MET-based calorie estimates (calories = MET x body weight in kg x hours). The usage examples are slash commands:

/fitness trend 3months

Others are /fitness analysis progress running and /fitness analysis correlation blood_pressure.

It is for people tracking their own workouts, and for health-tracking setups that already keep exercise, blood pressure and glucose data and want trends and correlations from them, with clear limits that keep it from acting as medical advice.