Imagyx

Find any image on your computer with natural language.

Download · How it works · Developer setup

Demo

Find a purple sunset, a red texture, a person near the sea, or just part of a filename — without organizing everything first.

IMAGYX-demo.1.webm

Why Imagyx?

• 🔎 Search how you think — describe an image naturally, search by filename, or combine both.
• ✨ Find similar images — pick any image and discover visually similar shots in your library.
• ⚡ Spotlight from anywhere — open the global search overlay, type, and act without opening the full app.
• 🖼️ Preview instantly — browse images full-size, inspect dimensions and tags, copy them, or reveal them in Explorer/Finder.
• 🔄 Convert in a click — create a new copy as AVIF, WebP, PNG, JPG, or ICO directly from Spotlight.
• 📁 Keep your library current — Imagyx watches followed folders and updates its index as images change.
• 🚀 Fast by design — progressive results, cached thumbnails, background indexing, and a lightweight idle footprint.
• 🎨 Made for your desktop — keyboard-first controls, light and dark themes, and 10 interface languages.

Download

Download Imagyx from GitHub Releases →

Windows and macOS builds are published manually on the Releases page.

How it works

1 · Add a folder 📁	2 · Let it index ⚙️	3 · Search anywhere 🔎
Choose a folder containing images. Imagyx discovers its files and prepares the local index.	The first pass creates metadata and the semantic index. Progress stays visible, and you can keep using the app.	Search in the app, or open Spotlight with Ctrl + 9 by default and act on a result immediately.

The shortcut is configurable in Settings. New and changed images are detected automatically in followed folders.

Private by default

Your images never leave your computer. Search, indexing, thumbnails, and semantic matching all run locally; Imagyx does not upload your library to a search service. Original files are never moved, renamed, or changed.

The first setup may download the local AI model files it needs. Once cached, normal image search remains local.

Contributing

Issues, ideas, bug reports, and pull requests are welcome. Development setup is documented separately for each supported platform:

• Windows development guide
• macOS development guide
• Developer documentation index

Developer information

Quick start

Requirements:

• Node.js 22.12 or newer;
• Rust 1.88 or newer;
• the native build dependencies documented in the platform guides.

npm install
npm run tauri dev

Local validation

npm run typecheck
npm test
npm run build
cargo fmt --manifest-path src-tauri/Cargo.toml --all -- --check
cargo check --manifest-path src-tauri/Cargo.toml
cargo test --manifest-path src-tauri/Cargo.toml
cargo clippy --manifest-path src-tauri/Cargo.toml --all-targets

Implementation overview

The desktop shell is built with Tauri and Rust. The interface uses Vue and TypeScript. Metadata, text search, and persisted vectors are stored locally with SQLite. Semantic search uses MobileCLIP-S0 through Transformers.js, with WebGPU acceleration when available and a WASM fallback.

The Rust backend is divided by responsibility under src-tauri/src/, including commands, database access, indexing, fuzzy search, tracing, and the in-memory vector store.

Release workflows and release publication are intentionally manual. Build artifacts can be produced with:

npm run tauri build

Additional performance and backend notes are available in docs/RUST_PERFORMANCE_PLAN.md.