YggPaperDebugger/paperdebugger
↗ GitHubA Plugin-Based Multi-Agent System for In-Editor Academic Writing, Review, and Editing
1,433
Stars
68
Forks
9
Watchers
14
Open Issues
Safety Rating A
No hardcoded secrets, malicious code patterns, suspicious dependencies, or prompt injection attempts were detected in the repository content. The project is a well-documented open source academic tool published on the Chrome Web Store under AGPL-3.0, with a clear architecture, community channels, and an associated arXiv paper. No red flags were identified.
ℹAI-assisted review, not a professional security audit.
AI Analysis
PaperDebugger is a plugin-based multi-agent AI system delivered as a Chrome browser extension that integrates directly with Overleaf, the online LaTeX editor. It provides an in-editor academic writing assistant powered by a custom MCP-based orchestration engine (XtraMCP) that simulates the full academic workflow: Research → Critique → Revision. The backend is written in Go with a microservices architecture (Gin/gRPC, MongoDB, OpenAI API, JWT auth) while the frontend is a TypeScript browser extension. It supports multi-step reasoning, reviewer-style critique, citation verification, and structured revision passes without leaving the Overleaf editor.
Use Cases
- AI-assisted academic paper writing, editing, and review directly inside Overleaf
- Multi-agent orchestration for literature-grounded research, AI-conference-style critique, and citation verification
- One-click AI suggestion insertion into LaTeX documents
- Automated comment generation and insertion into research papers
- Self-hosted backend deployment for institutional or privacy-sensitive academic use
Tags
Project Connections
CoPaw
→CoPaw provides a general-purpose multi-agent framework with MCP client support; PaperDebugger's custom XtraMCP orchestration engine could be integrated or compared with CoPaw's agent coordination layer for academic workflows.
claude-scientific-skills
→Claude Scientific Skills provides a library of research-domain agent skills (including scientific communication) that could augment PaperDebugger's academic writing and review pipeline with domain-specific expertise.
reader
→Reader's URL-to-LLM-friendly-text conversion capability could serve as a literature retrieval component feeding into PaperDebugger's research phase for citation verification and literature grounding.
zeroleaks
→ZeroLeaks tests LLM-based systems for prompt injection vulnerabilities; its evaluation methodology is relevant to assessing the security posture of PaperDebugger's multi-agent MCP orchestration engine.
Scrapling
→Scrapling's MCP server and adaptive web scraping capabilities could support PaperDebugger's literature research phase by programmatically retrieving academic content from web sources.