paperdebugger

PaperDebugger is an MCP-based backend that powers an AI-assisted academic writing workflow. The server orchestrates multi-step reasoning, reviewer-style critique, and structured revision passes to help researchers debug and improve their papers, with OpenAI integration for natural language capabilities and a Gin-based HTTP API served via a Go 1.24+ stack. It is designed to be self-hosted behind the extension, enabling the extension to communicate with the backend to fetch intelligent suggestions, in-editor critiques, and structured revisions without exposing raw content to external services. You can deploy the backend using Docker and connect it to a MongoDB instance; the system authenticates via JWT and supports OAuth flows as needed.

To use PaperDebugger, run the MCP server in your environment (typically via Docker) and ensure the backend URL is configured in the PaperDebugger extension (for self-hosted setups). The extension will send requests to the backend for tasks such as literature-grounded critiques, AI-assisted revisions, and in-editor comments. The Multi-Agent Orchestration (XtraMCP) integration enables specialized reasoning passes like literature review, conference-style evaluation, citation verification, and domain-specific revision passes when configured.

Prerequisites\n- Docker (and Docker Compose if you plan to use a multi-container setup)\n- Optional: MongoDB instance (or a Dockerized MongoDB) for data storage\n- Access to OpenAI API (or compatible LLM provider) and credentials\n\nStep-by-step installation\n1) Install Docker: follow instructions at https://docs.docker.com/get-docker/ for your OS.\n\n2) Pull and run the PaperDebugger MCP image:\n\nbash\ndocker pull paperdebugger/paperdebugger\ndocker run -d --name paperdebugger-mcp \ -p 8080:8080 \ -e MONGO_URI="mongodb://<user>:<pass>@mongo:27017/paperdebugger" \ -e OPENAI_API_KEY="<your-openai-api-key>" \ paperdebugger/paperdebugger\n\n\n3) Configure environment connections\n- MongoDB: set MONGO_URI to point to your database (host, port, credentials)\n- OpenAI API: set OPENAI_API_KEY or equivalent as required by the backend configuration\n\n4) Verify the service is running\nbash\ndocker ps | grep paperdebugger\n\n\n5) Point the PaperDebugger extension to your backend URL (e.g., http://localhost:8080) and test a few flows (AI critique, revision suggestions, and in-editor insertable comments).\n

Tips and notes:\n- Ensure your MongoDB instance is accessible by the MCP container; use a proper URI with credentials and TLS as needed.\n- Secure your OpenAI API key and rotate it periodically.\n- If you use self-hosted endpoints, you may need to configure CORS and JWT/OAuth settings to match the extension.\n- The MCP engine supports multi-agent workflows; enable XtraMCP modules or custom prompts to tailor literature-grounded reviews, conference-style evaluation, and domain-specific revisions.\n- Check logs if you encounter connection errors between the extension and backend; verify endpoint URL, network access, and firewall rules.\n