claude-scholar
Personal AI CLI configuration for academic research & software development. Supports Claude Code, OpenCode, and Codex CLI — covering the full research lifecycle from ideation to publication.
claude mcp add --transport stdio galaxy-dawn-claude-scholar npx -y galaxy-dawn/claude-scholar \ --env CLAUDE_SCHOLAR_CONFIG="path/to/config or placeholder"
How to use
Claude Scholar MCP is a configurable bridge that lets you deploy the Claude Code-based scholar configuration as an MCP server. It packages a rich set of skills, agents, and workflow commands to manage academic research tasks (like ideation, literature review, Zotero integration) and software development processes. Once the MCP server is running, you can invoke commands and skills via the CLI/agent interfaces defined in the project, such as research-ideation, literature-reviewer, zotero-review, and zotero-notes, to orchestrate Zotero-backed literature workflows, import items by DOI, attach PDFs, and generate review material. It also includes project management and coding-workflow commands to support end-to-end research and software development lifecycles, with cross-platform hooks and security rules. Use the provided commands to start, plan, and execute multi-step tasks, then inspect results (generated documents, structured notes, and organized Zotero collections) from within the MCP environment or exported artifacts.
To interact with the server, run the MCP client that corresponds to the installed package (via npx). Typical usage involves starting the server, then issuing commands like /research-init "topic" to begin a Zotero-backed literature workflow, /zotero-review "collection" to analyze a Zotero collection, or /zotero-notes "collection" to generate reading notes. The system is designed to guide you through ideation, planning, code review, and experiment tracking, with structured outputs such as research proposals, literature reviews, and BibTeX exports.
How to install
Prerequisites
- Node.js (recommended LTS) and npm installed on your system
- Git (optional for cloning or for versioned setup)
- Internet access to install the MCP server package
Installation steps
-
Install Node.js and npm if you don’t have them already. Visit https://nodejs.org and follow the installation instructions for your OS.
-
Install the Claude Scholar MCP server via npx (no global install required):
npm --version node --version
npx -y galaxy-dawn/claude-scholar
This will download and run the MCP server package for Claude Scholar with default configuration.
-
If you prefer to install and manage locally, clone the repository (or fetch the package from npm) and run the server using the provided entry point as documented in the repository. Follow the repository’s MCP_SETUP.md or setup.sh guidance for merging with existing Claude configurations.
-
Optional: run setup scripts to integrate with existing ~/.claude configurations, backup settings.json, and merge mcpServers/plugins as described in the changelog and installation notes.
-
Verify the server is running by checking the console for the MCP-ready message and try invoking a basic command such as /plan or /research-init "topic" to confirm proper operation.
Additional notes
Tips and caveats:
- If you are upgrading from an older setup, use the setup.sh installer to safely merge into ~/.claude and back up existing settings.json before applying changes.
- The project supports cross-platform Node.js; ensure your environment PATH includes Node executables used by npx.
- Some features rely on external services (Zotero, DOI lookups, PDF full-text retrieval). Make sure network access is available and authentication tokens (if any) are configured via environment variables or configuration files.
- For OpenCode, Codex CLI, or OpenCode alternatives, switch branches as documented to enable respective configuration files (CLAUDE.md vs CLAUDE.zh-CN.md).
- If you encounter issues starting the MCP, check the environment variable CLAUDE_SCHOLAR_CONFIG (set to a path or placeholder) and ensure the config file is accessible by the process.
- Typical outputs include generated documents (literature-review.md, research-proposal.md) and exported collections; ensure your workspace has write permissions where you run the MCP server.
Related MCP Servers
everything-claude-code
The agent harness performance optimization system. Skills, instincts, memory, security, and research-first development for Claude Code, Codex, Cowork, and beyond.
deepcontext
DeepContext is an MCP server that adds symbol-aware semantic search to Claude Code, Codex CLI, and other agents for faster, smarter context on large codebases.
sandboxed.sh
Self-hosted orchestrator for AI autonomous agents. Run Claude Code & Open Code in isolated linux workspaces. Manage your skills, configs and encrypted secrets with a git repo.
ask-user-questions
Better 'AskUserQuestion' - A lightweight MCP server/OpenCode plugin/Agent Skills + CLI tool that allows your LLMs ask questions to you. Be the human in the human-in-the-loop!
talkito
TalkiTo lets developers interact with AI systems through speech across multiple channels (terminal, API, phone). It can be used as both a command-line tool and a Python library.
vibe-check
Stop AI coding disasters before they cost you weeks. Real-time anti-pattern detection for vibe coders who love AI tools but need a safety net to avoid expensive overengineering traps.