LessAI is a desktop workbench designed to streamline Chinese text rewriting by introducing a reviewable, rollback-able, and write-back workflow. It targets users who need precise control over AI-assisted text modifications, offering a structured process that enhances traceability and efficiency compared to traditional editing methods.

How It Works

This application leverages Tauri 2 to provide a desktop experience. Users import text files (.txt, .md), which LessAI segments based on configurable granularity (sentence, phrase, or paragraph). It then invokes OpenAI-compatible APIs to generate rewrite suggestions for each segment. These suggestions are presented chronologically in a timeline view, highlighting differences (diffs) between original and revised text. Users can apply, ignore, or delete suggestions, with support for resuming interrupted sessions and exporting final output or overwriting the source file directly.

Quick Start & Requirements

• Installation: Recommended installation via GitHub Releases for Windows, macOS, and Linux. Source builds require Node.js 20+, pnpm 10+, and Rust stable.
• Prerequisites: Configuration of an OpenAI-compatible API endpoint (Base URL), API Key, and Model name is mandatory. System-specific Tauri prerequisites apply (e.g., WebView2 on Windows).
• Links: GitHub Releases: https://github.com/GTJasonMK/lessAI/releases. Tauri prerequisites: https://v2.tauri.app/start/prerequisites/.

Highlighted Details

• Granular text segmentation options (small sentence, whole sentence, paragraph).
• Timeline view displaying diffs for each modification suggestion, enabling detailed review.
• "Breakpoint resume" functionality allows interrupted rewriting sessions to be continued.
• Supports both manual, segment-by-segment generation and automatic batch processing.
• Includes built-in and customizable prompt templates for various writing scenarios.
• Direct "write back" feature to overwrite the original file with applied changes, followed by session cleanup.

Maintenance & Community

The project utilizes GitHub Actions for automated releases triggered by version tags. It benefits from community feedback and exchange, particularly mentioning the "linuxdo (linux.do) community".

Licensing & Compatibility

The project is released under the MIT License, which is permissive for commercial use and integration into closed-source projects.

Limitations & Caveats

LessAI does not provide its own model services; users must supply their own OpenAI-compatible API endpoint and credentials. API keys are stored in plaintext within the application's configuration file (settings.json), necessitating careful management of the application data directory to prevent exposure. The effectiveness of the rewriting heavily relies on the chosen AI model and prompt configurations.