FootFix

Batch image resizer with LLM-powered alt text generation. A desktop app for editorial teams and content managers who need web-ready images without cloud uploads or subscriptions.

What It Does

• Batch resize images to a max dimension with automatic quality optimization
• Target file size via binary-search compression (e.g. "fit under 500 KB")
• Custom filename templates with tokens like {name}_{width}x{height}
• Save/load/share presets as .json files for team-wide consistency
• Generate alt text using OpenAI, Anthropic, or Google LLMs
• 100% local processing — images never leave your machine (only LLM API calls for alt text)

Tech Stack

Getting Started

# Install dependencies
npm install

# Development
npx electron-vite dev

# Build
npx electron-vite build

# Package (Windows)
npm run package

Project Structure

src/
  main/              # Electron main process
    index.ts           Entry point, window creation
    settings-store.ts  Persisted settings + encrypted API keys
    image-processor.ts Sharp pipeline + binary-search quality
    preset-manager.ts  Preset CRUD + file dialogs
    llm-service.ts     Alt text generation via multi-llm-ts
    filename-template.ts  Template rendering
    file-utils.ts      Folder selection dialog
  preload/           # Context bridge (14 typed IPC methods)
  shared/            # Types + IPC channel constants
  renderer/src/      # Svelte 5 UI
    App.svelte         App shell with sidebar navigation
    views/             3 views (Process, Results, Settings)
    lib/components/    10 UI components
    lib/stores/        3 reactive stores (job, settings, preset)
    lib/utils/         Formatters + IPC helper

Architecture

Components call store actions. Stores call window.api. The preload bridge maps to ipcMain.handle() handlers in the main process. All 14 IPC channels use typed constants from src/shared/ipc-channels.ts.

See IMPLEMENTATION.md for detailed architecture and phase status. See PRODUCT.md for user stories and product requirements.