SlothSort v0.1.1

SlothSort v0.1.1 — Windows Patch Release Candidate

SlothSort v0.1.1 is a security and release-hygiene patch release candidate for Windows x64.

Highlights

• Blocks execution and undo through NTFS junctions, symlinks, and other reparse-point paths that resolve outside the selected root.
• Requires a scan-time SHA-256 source fingerprint before an approved file can be moved.
• Rejects approval before analysis completes and validates edited proposals, including confidence values, in the main process.
• Prevents destination replacement when a competing file appears between preflight and execution.
• Hardens cross-volume copy, verification, finalization, and source-removal failure cleanup so the source is preserved and temporary destinations are removed or rolled back.
• Removes the Google Fonts dependency and keeps packaged renderer requests local-only during the validated workflow.
• npm test covers 162 deterministic checks, including the 10-test test:hardening suite. The separate npm run test:windows-hardening command covers 5 Windows/NTFS cross-volume hardening checks.
• Corrects the Electron Builder allowlist so previous dist/package and dist/release outputs are not recursively embedded in app.asar.

Validation evidence

• Windows 11 build 10.0.26300.9711, AMD64.
• Node.js v26.10.0, npm 11.19.1.
• Ollama 0.40.6 with gemma3:4b installed locally.
• C: and D: confirmed as NTFS volumes.
• Windows npm test: 162 deterministic checks passed.
• Windows npm run test:windows-hardening: 5/5 checks passed.
• Fresh packaged workflow passed: launch, local model detection, synthetic scan, inference, edit, approval, move, history, undo, restart, and renderer resource audit.
• Extracted ZIP executable launched successfully.

Candidate artifact

• Filename: SlothSort-v0.1.1-win-x64.zip
• Executable: SlothSort.exe
• Executable size: 246,588,416 bytes
• Executable SHA-256: eb2ddc7eae02ed42757a29f59c0e8280b6a4ee95b0f750eb0f42f15c91859c82
• ZIP size: 158,411,658 bytes
• ZIP SHA-256: 9749092838b08d14641ecb483d893fd53224645a90cc54844b4959494b8b3f42

Known limitations

• The Windows executable is unsigned and may trigger the SmartScreen “Unrecognized app” warning. Select More info → Run anyway if you trust the downloaded artifact.
• The model is not bundled. An initial internet connection is required to install Ollama and run ollama pull gemma3:4b; inference itself uses local loopback only.
• The packaged workflow was validated on the development Windows 11 machine. A clean third-party Windows machine without developer tools was not available for this run.
• CPU inference remains slow on consumer hardware, approximately 20 seconds per file on the measured Ryzen 5 5500U system.
• The shared WSL2/Linux workspace is not currently repeatable after Windows-native dependencies were installed; earlier Linux/WSL checks passed before that switch, and the Windows run is authoritative for this Windows candidate.

Release status

This is a release candidate only. No Git commit, tag, push, GitHub release, or DEV article publication was performed as part of this preparation.

SlothSort v0.1.0 — Windows Desktop Release

SlothSort v0.1.0 — Windows Desktop Release

SlothSort is a privacy-first desktop file organizer that uses a locally running AI model to suggest file categories, destination folders, and clearer filenames. You review the proposals before changes are executed.

Download

Download SlothSort-v0.1.0-win-x64.zip, extract it, and launch SlothSort.exe.

Requirements

• Windows x64
• [Ollama](https://ollama.com/) installed and running
• The gemma3:4b model downloaded through Ollama

Run ollama pull gemma3:4b in PowerShell to download the model before using AI-powered organization.

Note: The model weights are not included in the ZIP and must be downloaded separately.

Highlights

• Local AI-powered file organization suggestions
• Human review before file operations
• Filename and destination validation
• Operation history and supported undo workflows
• Privacy-oriented scanning and sensitive-file exclusions
• No configured cloud AI fallback in the tested application

Performance and limitations

On an AMD Ryzen 5 5500U test system, observed AI inference took approximately 19.6–20.7 seconds per file. Performance varies with hardware and model state.

This release has been tested on a development machine; a clean-machine installation test has not been completed. The Windows executable is unsigned, so Windows SmartScreen may show a warning.

Verify the download

SHA-256:

0C97D4AC58323A13D08263027896EBEEDDFC0245D1E2573CD6327A25F5DF41AF

Project and feedback

Source code, documentation, and tests: https://github.com/Pranjal-debug/slothsort

This is the initial public release. Feedback, bug reports, and contributions are welcome.