Sphinx

AI-assisted metadata for stock photo and video contributors.
Free and open source, for Windows, macOS and Linux.

Website · Download · Quick start

Sphinx is a free desktop app for Windows, macOS and Linux. Drop in your photos and videos, and it writes titles, descriptions and keywords for you, shaped to each stock site's rules. It embeds them into your files and uploads them to your contributor accounts.

_{Demo with sample data.}

Features

• AI analysis: describes each image or video with a vision model of your choice. Cloud options are OpenAI, Google Gemini and Anthropic Claude; a local Ollama model keeps everything on your machine.
• Per-site metadata: titles, descriptions and keyword counts are trimmed to each site's limits (Adobe Stock, Shutterstock and others, plus custom site profiles). Edits can be saved separately for each site.
• Keyword editor: keywords are scored and colour-coded by strength, and can be reordered, removed, restored or enriched with real search-demand data from Shutterstock and Adobe Stock.
• Video support: extracts keyframes for analysis, with optional speech transcription.
• Embedding and export: writes IPTC/XMP metadata into your files with exiftool, or exports each site's CSV upload format.
• Upload: delivers files over SFTP or FTPS straight to your contributor accounts.
• Batch jobs: queue analysis, embedding and uploads for many files and track them on the Activity screen.

Install

Download the latest installer from the Releases page:

OS	File
Windows 10/11	-setup.exe or .msi installer, or _portable.exe to run without installing
macOS (Apple Silicon / Intel)	.dmg for your chip (aarch64 / x64)
Linux	.AppImage, .deb or .rpm

No-install options: the Windows _portable.exe and the Linux .AppImage are single files you can run directly. On Linux, run chmod +x on the AppImage first. The portable exe needs Microsoft's WebView2 runtime, which Windows 11 already includes; on older Windows 10, use the installer, which sets it up. Either way, Sphinx stores its library in the data folder listed below, not next to the program.

First launch warning: the installers aren't signed with a publisher certificate yet, so your OS will warn you the first time:

• Windows: if SmartScreen says "Windows protected your PC", click More info → Run anyway.

• macOS: drag Sphinx to Applications and open it. When macOS says it can't verify the app, click Done, then go to System Settings → Privacy & Security, scroll down and click Open Anyway next to the Sphinx message. Confirm with your password, then click Open Anyway again. You only need to do this once.

  If macOS instead says the app "is damaged", run this in Terminal, then open the app again: xattr -dr com.apple.quarantine /Applications/Sphinx.app

Also install these two free tools

Sphinx relies on them but doesn't bundle them:

Tool	Used for	Get it
ExifTool	Writing metadata into your files	https://exiftool.org (macOS: brew install exiftool, Linux: apt install libimage-exiftool-perl)
FFmpeg	Reading video keyframes and audio	https://ffmpeg.org (macOS: brew install ffmpeg, Linux: apt install ffmpeg)

Put both on your PATH, or set their locations under Connections → Local tools. The About screen shows whether Sphinx can find them.

Quick start

1. Connections: choose an AI provider and paste its API key, or point Sphinx at a local Ollama server. Optionally add SFTP/FTPS logins for your stock sites, and keyword-API keys for enrichment.
2. Import: add files or a folder. Sphinx can also watch a folder for new files.
3. Library: run the pipeline on new items (analyze, then generate metadata), then review the results. Open any item in the editor to adjust its title, description and keywords for each site.
4. Embed the metadata into the files, then upload them or export a CSV for the site.

Privacy and security

• No account, server, telemetry or tracking. Sphinx runs entirely on your computer.
• Passwords and API keys go into your operating system's credential store (Windows Credential Manager, macOS Keychain, or GNOME Keyring / KWallet on Linux), never into Sphinx's database.
• Your media only leaves your machine when you choose a service:
  • A cloud AI provider receives the images or keyframes it analyses, under that provider's own terms. Use Ollama to keep media local.
  • Your SFTP/FTPS servers receive the files you upload.

See SECURITY.md for details and how to report a vulnerability.

Where your data lives

The library database (sphinx.db) is stored in:

OS	Location
Windows	%APPDATA%\com.thusitha.sphinx\
macOS	~/Library/Application Support/com.thusitha.sphinx/
Linux	~/.local/share/com.thusitha.sphinx/

Deleting that folder resets Sphinx. Saved credentials are listed in your OS credential store under sphinx-sftp and sphinx-api-keys. Sphinx never modifies or deletes your original files, except to write metadata into them when you embed.

Building from source

Requirements:

• Node.js 18+
• Rust (stable)
• The Tauri prerequisites for your OS: https://tauri.app/start/prerequisites/
  • On Linux, also install libdbus-1-dev and pkg-config (needed for the credential store).

Then:

npm install
npm run tauri dev      # run with hot reload
npm run tauri build    # build an installer for your OS

Run the engine's unit tests (no network or windowing toolkit needed):

cd src-tauri/crates/core
cargo test

Project layout

• src/: React + TypeScript frontend (Vite).
  • Screens live in src/routes/.
  • Shared UI lives in src/components/.
  • App state is in src/state/AppContext.tsx.
• src-tauri/: the Tauri v2 desktop shell. It exposes the engine to the frontend as commands in src-tauri/src/lib.rs.
• src-tauri/crates/core/ (sphinx-core): the engine, a plain Rust library with no GUI dependency.
  • db.rs: SQLite schema and numbered migrations.
  • ingest.rs / watch.rs: importing, BLAKE3 deduplication and folder watching.
  • analysis/: vision providers (OpenAI, Gemini, Anthropic, Ollama).
  • metadata/: per-site limit profiles.
  • embed/: exiftool embedding and CSV export.
  • keywords/: Shutterstock and Adobe Stock enrichment.
  • video/ and transcribe/: keyframes and speech-to-text.
  • upload/: SFTP (russh) and FTPS (suppaftp).
  • secrets.rs: OS credential-store access.

Releases are built by .github/workflows/release.yml when a v* tag is pushed.

Contributing

Issues and pull requests are welcome. See CONTRIBUTING.md.

Support the project

If Sphinx saves you time, you can buy me a coffee ☕

License

MIT © Thusitha Jayasundara