A featherweight, always-on file converter that lives in your menu bar. Drop any file onto the floating window, or select files and hit a hotkey, and get instant format conversion without opening a browser or bulky software.
- Floating drop zone -- a small, always-on-top window accepts drag-and-drop from Finder or any file manager
- Global hotkey -- cmd + shift + Space toggles the window; conversion options appear immediately
- Zero-friction UX -- one click to pick the output format, conversion starts instantly
- Broad format support
- Images: WebP, PNG, JPEG, AVIF, GIF, BMP, TIFF, HEIC/HEIF
- Video: MP4, MOV, MKV, WebM, AVI, GIF (video-to-GIF and GIF-to-video)
- Audio: MP3, AAC, FLAC, OGG, WAV, OPUS, M4A
- Data: JSON, YAML, TOML, CSV
- Documents: Markdown, plain text, HTML, LaTeX, Typst, PDF, EPUB, MOBI (via pandoc + pymupdf4llm/marker-pdf/Calibre)
- PDF merge -- combine multiple PDF files into one in a dedicated merge mode
- Clipboard integration -- paste (⌘V) to load clipboard content for conversion; result written back to clipboard or saved to disk (configurable)
- Local LLM post-processing -- optional Ollama integration cleans up extracted PDF text (configurable model and endpoint)
- Output next to source -- converted file lands in the same folder as the input (configurable)
- Batch conversion -- drop multiple files at once; common formats shown in a shared toolbar
- Progress indicator -- lightweight inline progress bar per file, no modal dialogs
- Click to reveal -- success label opens the output folder in Finder
- Auto-update -- checks for new releases on startup; installs in the background
- Settings panel -- configure output folder, JPEG/AVIF quality, concurrency, PDF conversion engine, and local LLM
| Layer | Technology |
|---|---|
| App shell | Tauri v2 |
| Backend logic | Rust |
| Image processing | image crate + ravif for AVIF + macOS sips for HEIC |
| Video / audio | System ffmpeg (auto-installed via Homebrew if missing) |
| Document conversion | System pandoc (auto-installed via Homebrew / winget / apt if missing) |
| PDF ↔ EPUB / HTML / MD | pymupdf4llm (auto-installed via pipx); optional ML engine via marker-pdf |
| MOBI conversion | System ebook-convert / Calibre (auto-installed via Homebrew cask / winget / apt) |
| Local LLM | Ollama (optional); cleans up PDF extraction output |
| Data serialization | serde_json, serde_yaml, toml, csv crates |
| Frontend UI | TypeScript + Vite (plain DOM, no framework) |
| Global hotkeys | tauri-plugin-global-shortcut |
| System tray | Tauri built-in tray API |
| Auto-update | tauri-plugin-updater |
swift-shifter/
├── swift-shifter/ # Rust/Tauri workspace member
│ ├── src/
│ │ ├── main.rs # Tauri app entry point, window + tray + menu setup
│ │ ├── config.rs # Persisted settings (quality, concurrency)
│ │ ├── clipboard.rs # Clipboard read/write (paste-to-convert, copy result back)
│ │ ├── hotkey.rs # Global shortcut registration
│ │ ├── tray.rs # System tray icon and menu
│ │ └── converter/
│ │ ├── mod.rs # Dispatcher: routes files to the right converter
│ │ ├── image.rs # image crate + ravif (AVIF) + sips (HEIC)
│ │ ├── media.rs # ffmpeg subprocess wrapper
│ │ ├── data.rs # JSON/YAML/TOML/CSV converters
│ │ └── document/ # pandoc + pymupdf4llm + marker-pdf + Calibre (md/txt/html/pdf/epub/mobi/tex/typst)
│ │ ├── mod.rs
│ │ ├── binaries.rs # dependency detection and auto-install
│ │ ├── conversion.rs # conversion logic per format
│ │ ├── merge.rs # PDF merge (combine multiple PDFs into one)
│ │ ├── llm.rs # Ollama LLM post-processing
│ │ ├── types.rs
│ │ └── utils.rs
│ ├── capabilities/
│ │ └── default.json # Tauri v2 permission grants
│ ├── Cargo.toml
│ └── tauri.conf.json
├── ui/ # Frontend (loaded by Tauri webview)
│ ├── index.html # Main drop-zone window
│ ├── settings.html # Preferences window
│ └── src/
│ ├── main.ts # Drag-drop, file picker, conversion UI logic
│ ├── settings.ts # Settings panel logic
│ ├── style.css # Main window styles (adaptive light/dark)
│ ├── settings.css # Settings panel styles
│ └── tokens.css # Shared design tokens
├── .github/
│ └── workflows/
│ ├── build.yml # Compile check on macOS, Windows, Ubuntu, Fedora, Arch
│ ├── tag.yml # Auto-creates v* tag after build passes on main
│ └── release.yml # Builds signed installers and publishes GitHub release
├── scripts/
│ ├── bump-version.sh # Syncs version across package.json, Cargo.toml, tauri.conf.json
│ └── build-icons.sh
├── vite.config.ts
├── tsconfig.json
└── package.json
| Tool | Version | Notes |
|---|---|---|
| Rust | stable (≥ 1.78) | via rustup |
| Node.js | ≥ 24 | for Tauri CLI and Vite |
| ffmpeg | ≥ 6 | for video/audio; auto-installed via brew if missing |
| pandoc | any | for document conversion; auto-installed via brew / winget / apt |
| pymupdf4llm | any | for PDF extraction; auto-installed via pipx if missing |
| Calibre | any | for MOBI conversion; auto-installed via brew / winget / apt |
| Ollama | any | optional; for LLM post-processing of PDF text (user-installed) |
# Clone
git clone https://github.com/mtmatt/swift-shifter.git
cd swift-shifter
# Install dependencies (Tauri CLI + Vite + TypeScript)
npm install
# Run in dev mode (Vite hot-reload + Rust watch)
npm run tauri -- dev
# Build release binary
npm run tauri -- buildThe release .app lands in swift-shifter/target/release/bundle/.
./scripts/bump-version.sh 0.2.0
git push origin mainThis syncs package.json, Cargo.toml, and tauri.conf.json in one step. Pushing to main triggers the CI pipeline, which automatically creates the v0.2.0 tag and kicks off the release build.
| Workflow | Trigger | What it does |
|---|---|---|
build.yml |
Push to main, PRs |
Compile check on macOS, Windows, Ubuntu, Fedora, Arch |
tag.yml |
build.yml passes on main |
Verifies version files agree, creates v{version} tag |
release.yml |
v* tag pushed |
Builds signed installers on all platforms, publishes GitHub draft release |
- Core Tauri shell + system tray
- Drop zone window with drag-and-drop
- Global hotkey (⌘⇧Space) → file picker fallback
- Image conversion (PNG/WebP/JPEG/AVIF/HEIC)
- Data conversion (JSON/YAML/TOML/CSV)
- Video/audio via ffmpeg subprocess
- Progress events streamed to UI
- Output folder reveal
- Batch conversion with concurrency
- Config file + settings panel
- Handle ffmpeg and brew installation automatically
- Auto-update via Tauri updater plugin
- Windows and Linux support
- CI/CD pipeline with automatic tagging and signed releases
- Document conversion via pandoc
- Images to PDF
- EPUB <-> PDF
- MOBI <-> PDF/EPUB/HTML/MD
- PDF/EPUB/MOBI -> HTML and Markdown
- Local LLM post-processing for PDF conversions (Ollama)
- PDF merge (combine multiple PDFs into one)
- M4A audio format support
- Clipboard integration (paste to convert; copy result back to clipboard)
- Clipboard Integration: Convert image/text directly from clipboard to file and vice versa
- Watched Folders: Designate folders for automatic background conversion upon file drop
- Smart Presets: Create custom "Action Chains" (e.g., Convert to WebP + Resize to 1200px + Strip Metadata)
- Post-processing hooks: Run custom shell commands or scripts after a successful conversion
- OCR conversion (PDF/image → txt, md, tex, typst)
- AI Background Removal: Remove image backgrounds locally using RMBG/ONNX models
- Super Resolution: Local AI upscaling for low-res images
- Privacy Shield: Auto-detect and blur faces or sensitive information (PII) before conversion
- Offline LLM Support: Ollama integration for local LLM post-processing of PDF text
- Deeper LLM integration: document translation and summarization via local models
- Video/Audio to Markdown, text, or SRT subtitle files using local Whisper.cpp.
- Swift Shifter indexes the metadata and AI-generated tags of your recently converted files, allowing you to hit the hotkey and type "that picture of a dog I converted yesterday" to locate it.
- Drop an image with an aspect ratio constraint (e.g., "convert to 16:9") and use local Stable Diffusion/Burn to intelligently fill the missing background rather than cropping or stretching.
- Local AI splitting of audio tracks into isolated vocals, drums, bass, and instruments (via Demucs/ONNX integration).
- Drop an image or video file and output a CSS, JSON, or Tailwind config file containing the dominant color palette and hex codes.
- Convert between standard 3D formats (OBJ, FBX, STL) to web and AR-ready formats (GLTF, GLB, USDZ).
- Drop a
.rs,.ts, or.pyfile and generate a beautifully syntax-highlighted, customized image for social sharing (similar to Carbon). - Auto-detect subjects in landscape videos and convert them to portrait (9:16) for social media, keeping the primary subject in the frame.
- Run SVGs through a Rust-based optimization pipeline (like
usvg) to strip bloat and standardise paths before dropping them into a project.
- Automatically pause or throttle heavy operations (like video encoding or local LLM tasks) when a laptop switches to battery power or enters low-power mode.
- Register Swift Shifter into the native Windows Explorer and macOS Finder right-click context menus for users who prefer mouse-driven workflows over drag-and-drop.
- Convert a file and immediately generate a temporary local-network QR code or link to securely download the output on a mobile device without touching the cloud.
- Auto-detect Apple Silicon (Metal), Nvidia (CUDA), or Intel (QuickSync) hardware on first boot and dynamically compile or select the optimal ffmpeg and ONNX backend configurations.
- Specifically dither and contrast-boost PDFs and images for optimal reading on Remarkable, Kindle, or Boox devices.
- Invisibly embed a text file or JSON payload directly into the pixel data of an image output.
Apache License Version 2.0