Organize, search and chat with your files using AI that runs on your own computer.
Download · Documentation · Website · Changelog
The tour at 2× speed. Watch the full-quality video.
StratoSort reads what is inside your files (documents, spreadsheets, images, scans, audio and video), suggests a clear name and the right folder for each one, and lets you search and ask questions across all of them. The AI runs on your machine: there is no account, no cloud service and no telemetry.
- Organizes by content. Suggests names and folders from what a file contains, not its filename. You review every suggestion before anything moves, and every move can be undone.
- Smart Folders. Describe a folder in plain words ("invoices and receipts") and matching files are routed into it, including new files as they arrive.
- Search by meaning. Find the report about "how warming affects wildlife" even when those words are not in its name. Exact matches such as an invoice number still work.
- Chat with your documents. Ask questions and get answers built from your own files, with sources.
- See how files connect. Relate (a graph) and Atlas (a map) show related files, clusters and near-duplicates.
- Reads images, scans and recordings. Vision and OCR for photos, screenshots and scanned PDFs; local Whisper transcription for audio and video.
- Works with other AI tools. An opt-in local MCP server lets Claude Desktop, Cursor or Zed search your files without uploading them.
- Optional cloud AI. Bring your own Anthropic or OpenAI-compatible key for some tasks, with a monthly spend cap. Search and embeddings always stay local. See Cloud AI Providers.
- Download the installer for your system from GitHub Releases.
- Launch it and choose a model profile. StratoSort downloads its AI models once: about 7 GB for Base, or about 10 GB for Better Quality.
- Add a folder and review the suggestions.
No terminal, Python, Docker or API keys. Windows installers are Authenticode-signed and macOS builds are signed and notarized by Kennedy Applied Sciences, Inc. The Install Guide covers first-launch prompts and how to check the download's checksum.
| System | Supported | Installers |
|---|---|---|
| Windows | Windows 10 (22H2) or 11, x64 or arm64 | Installer or portable |
| macOS | macOS 13 (Ventura) or later, Apple Silicon or Intel | DMG or ZIP |
| Linux | Ubuntu 22.04 or later and derivatives, x64 or arm64 | AppImage or .deb |
| Memory | 8 GB RAM minimum; 16 GB RAM and an 8 GB GPU for Better Quality | |
| GPU | Optional. NVIDIA CUDA, Apple Metal or Vulkan; falls back to CPU |
Need it on a machine with no internet access? See Offline and air-gapped install.
- Analysis, search and chat run on your computer with AI models stored locally.
- No account, no analytics, no crash reporting to anyone.
- After the one-time model download, the only automatic connection is the update check, which you can turn off in Settings → Application → Background update checks. Every connection the app can make is listed in Network Transparency.
- The search index, including text extracted from your files, stays in the app's data folder on your machine.
See SECURITY.md to report a vulnerability.
Documents and office files, text and code, images, audio, video, archives, and 3D/design files: over 100 file types in all.
Full list
- Documents and office:
.pdf,.doc,.docx,.xlsx,.pptx,.odt,.ods,.odp,.epub,.eml,.msg,.kml,.kmz,.ipynb - Text, code and data:
.txt,.md,.rtf,.tex,.csv,.json,.xml,.yml,.yaml,.html,.htm,.css,.scss,.sass,.less,.js,.ts,.jsx,.tsx,.sql,.py,.java,.c,.cpp,.h,.rb,.go,.rs,.php,.sh,.bat,.ps1,.ini,.conf,.log,.scad. Files over the text size limit (50 MB by default) are analyzed from their first part. - Images:
.jpg,.jpeg,.png,.gif,.bmp,.webp,.svg,.tiff,.tif,.ico;.heicand.heifwhen the bundled FFmpeg can decode them, otherwise by filename and date - Audio (transcribed locally):
.mp3,.wav,.flac,.aac,.ogg,.oga,.m4a,.m4b,.opus,.wma,.aiff,.aif - Video (audio transcribed; keyframe descriptions optional):
.mp4,.mov,.mkv,.webm,.avi,.flv,.m4v,.wmv,.3gp,.mts,.m2ts - Archives:
.zip,.tar,.tgzand.gzare listed by content;.rar,.7zand.bz2are organized by name only - 3D and design:
.stl,.obj,.3mf,.gcode,.scad,.ai,.eps(metadata and text);.psdby filename and date only
Legacy .xls and .ppt files are organized by filename and date only; save them as .xlsx or
.pptx to have their contents analyzed.
The same docs are published at stratosort.com/docs.
Requires Node.js 20.19 or later (see .nvmrc).
git clone https://github.com/iLevyTate/StratoSortStack.git
cd StratoSortStack
npm ci
npm run devThe app opens with the same first-launch model setup as the installer. Electron 45, React 19, node-llama-cpp for local inference and Orama for search; see Architecture and Getting Started for the rest.
- Bugs: open a bug report
- Ideas: open an issue
- Code: read CONTRIBUTING.md, then open a pull request. Run
npm testbefore you do. - Testing a release: the Tester Guide explains what to check and how to report it.
Free for personal and other noncommercial use under the PolyForm Noncommercial License 1.0.0. Commercial use needs a paid license from Kennedy Applied Sciences, Inc.; see Commercial Licensing. Third-party components keep their own licenses (Third-Party Notices).
StratoSort Stack succeeds the original StratoSort and was inspired by llama-fs, Local-File-Organizer and ai-file-sorter.
© 2026 Kennedy Applied Sciences, Inc.