openGym

openGym is a workout tracker by GitHub user DuarteSantos8 for logging sessions and planning programs, with a web app (PWA) and native apps for iPhone, iPad, and Android. The mobile apps work offline on their own or sync with a server you host. The author writes in the README that "code, tests and documentation is drafted in Claude Code sessions." The AI coach and MCP are optional and use your own Anthropic, OpenAI, Gemini, or compatible API key; the coach can adjust suggestions based on your training log. By default it calls no language model and has no telemetry, and your data stays in your own directory.

Who it's for:Fitness enthusiasts who want to keep their training data in their own hands and don't mind setting up a server. If you just want an app for logging, the mobile version works on its own.

Before you start: The server runs with Docker Compose and opens on local port 8080. You can create a profile or use guest mode, and you sign in to your own server with a passkey. Weight, heart rate, progress photos, and HealthKit and Health Connect data can sync. Once you turn on the AI coach, your training data goes to the model service you configure, and cloud models may cost money. The app is free with no subscription or ads. The license has two parts: the code is AGPL-3.0, while the exercise photos and animations are licensed separately for use only in openGym and in unmodified, non-commercial forks that keep the notice; other uses need a separate media license.

Based on the project's public documentation. Not tested by AI Inspiration World.