Private Workout Log

Android-first Flutter app for logging gym workouts. Track exercises, body parts worked, sets and reps per session, dated. Offline-only, SQLite-backed, no account, no network.

What it is

A personal workout log built around two design choices most fitness apps don't make: it works fully offline, and it has no account, no cloud, no telemetry. Your data stays in a local SQLite database on the device, with manual JSON backup if you want it.

Single-developer Flutter project, published to Google Play (com.lukk.workoutlog), and intentionally kept narrow: log what you lifted, when, and which body parts you worked. No social features, no built-in exercise library to maintain, no premium tier.

Features

• Calendar-based session log; tap a date to see that day's sets
• Add exercises with primary and secondary body parts (custom list, no fixed library)
• Track sets and reps per exercise per session
• Local SQLite persistence — no account, no cloud, no telemetry
• Manual JSON backup and restore to external storage
• Configurable rest timer with a local notification when it fires
• Light / dark theme with custom background images
• Pure Riverpod state, providers in lib/presentation/providers/

Architecture

graph TB
    subgraph Presentation
        Pages[Pages]
        Widgets[Shared widgets]
        Providers[Riverpod providers]
    end
    subgraph Domain
        Models[Freezed value types]
    end
    subgraph Data
        DAOs[sqflite DAOs]
        Backup[BackupService]
        Alarm[AlarmService]
    end

    Pages --> Providers
    Widgets --> Providers
    Providers --> DAOs
    Providers --> Backup
    Providers --> Alarm
    DAOs --> SQLite[(SQLite on device)]
    Backup --> File[(backup.json on external storage)]
    Alarm --> FLN[flutter_local_notifications]

Loading

lib/ follows the standard three-layer split. Pages and widgets read and write through Riverpod providers; the providers wrap DAOs and services. Nothing in the presentation layer talks to SQLite or the notification plugin directly. The domain models in lib/domain/ are all freezed value types with generated equality and JSON.

Quick start

Clone the repo, then from the project root run the standard Flutter bring-up:

flutter pub get

Generate the freezed and json_serializable outputs (one-time, and again whenever a model changes):

dart run build_runner build

If a regen fails because stale generated files conflict, wipe them first:

dart run build_runner clean

(The old --delete-conflicting-outputs flag was removed in build_runner 2.15+.)

List attached devices and emulators:

flutter devices

Run on a specific device, where <deviceId> is the id printed by the previous command (e.g. emulator-5554):

flutter run -d <deviceId>

Add -v for verbose output or --release for a release-mode build.

If you hit weird build-cache issues:

flutter clean

Deployment

Release builds ship to the Google Play Store via a manual GitHub Actions workflow. See docs/DEPLOYMENT.md for the full procedure: how to trigger a release, the five GitHub secrets you need to configure (upload keystore + Google Play API service account), what the workflow actually does step-by-step, common failure modes and their fixes, and the manual flutter build appbundle path for when CI is unavailable. The first-time keystore-signing setup is in there too.

Docs

File	Covers
docs/DEPLOYMENT.md	Release process, signing model, GitHub secrets
docs/AGENT_TOOLING.md	How the project is set up for AI coding agents (Claude Code, Codex, etc.)
AGENTS.md	Agent-facing instructions and skill references
.agents/skills/	Project-scoped skills (code-formatter, openspec workflow, etc.)

License

Personal project, all rights reserved by Łukasz Sarna. No public license. The source is published for portfolio and agent-tooling demonstration; not intended for redistribution or repackaging.