MyFitnessPal without the data mining

What it does

SparkyFitness is a self-hosted health and fitness platform with a web frontend, backend API, and native iOS and Android apps. It tracks nutrition, exercise, hydration, sleep, fasting, mood, and body measurements — basically the full MyFitnessPal experience, except your data stays on infrastructure you actually control.

The interesting bit

The integration list is surprisingly deep: Apple Health, Google Health Connect, Fitbit, Garmin, Withings, Polar Flow, Strava, plus food databases like OpenFoodFacts and USDA. There’s even a beta AI chatbot (“SparkyAI”) that lets you log meals by uploading photos or typing natural-language descriptions. The project explicitly warns that auto-updating containers is not recommended — a refreshingly honest admission in the self-hosted space.

Key highlights

• Full stack: backend, web UI, and native mobile apps for iOS and Android
• Syncs with major fitness platforms and food databases; some integrations partially tested or untested
• Supports OIDC, TOTP, passkeys, and MFA out of the box
• Multiple user profiles with family access (beta)
• Light and dark themes, because obviously
• Cloud deployment option via PikaPods for the Docker-averse

Caveats

• AI chatbot, family access, and API documentation are all flagged as beta/untested
• The README warns of breaking changes between releases and explicitly discourages auto-updates
• Several integrations (Polar Flow, Hevy, Strava) are noted as partially tested or not tested at all

Verdict

Worth a look if you want MyFitnessPal-style tracking without the corporate data harvesting, and you’re comfortable self-hosting. Skip it if you need battle-tested reliability or a polished mobile experience right now — this is actively developed software with visible rough edges.