Summary Abnerla/AI_paper is a Windows-native, local-first AI academic writing assistant. It offers a unified platform for generating outlines, drafting content, reducing AI detection and plagiarism scores, polishing text, and correcting errors. Designed for researchers and students, it prioritizes data privacy by keeping operations and API keys local, enabling efficient, secure academic writing.

How It Works This Windows desktop app emphasizes local data storage and encrypted API key management. It integrates with over 20 AI providers (e.g., OpenAI, DeepSeek) and custom OpenAI-compatible endpoints. Core functions include AI content generation, text rewriting for AIGC detection/plagiarism, academic polishing, and error correction. AI model interactions require network access.

Quick Start & Requirements Download and run 纸研社.exe from Releases; no external dependencies. Windows 7+ (32/64-bit). Configure AI API keys in "API 配置". Network needed for AI features.

Highlighted Details

• AIGC Detection Reduction: Three rewrite intensities, per-paragraph AI scanning. Imports AIGC reports (DOCX/PDF) to flag risks.
• Plagiarism Reduction: Three rewrite levels (20-80%) with local duplicate detection. Supports importing plagiarism reports.
• Academic Polishing: Tailored modes (grammar, vocabulary, logic, comprehensive) for specific text types.
• Intelligent Error Correction: Addresses 8 error categories (grammar, format, citation, logic, AI traces) with severity levels and one-click fixing.
• Editor & Workflow: Rich text editor, .docx import/export, LaTeX conversion, auto-save, 200-operation history (diff/rollback), token usage tracking.

Maintenance & Community Under active development, soliciting feedback via GitHub Issues and email (773131249@qq.com). Acknowledgements list contributors. Donation support welcomed.

Licensing & Compatibility License not explicitly stated. Windows-only. AI model interactions require internet.

Limitations & Caveats Windows-only. Users must critically review all AI-generated content for accuracy and academic integrity before submission; direct use is discouraged. Over-reliance on AI is cautioned against.