mendeley
MCP server for Mendeley reference manager - search, retrieve, and manage your academic library from Claude and other MCP clients
claude mcp add --transport stdio pallaprolus-mendeley-mcp mendeley-mcp
How to use
The Mendeley MCP Server exposes your Mendeley reference library to MCP-compatible clients and LLM tools such as Claude Desktop and other MCP-enabled assistants. It supports functionalities like searching your library, browsing folders, retrieving full metadata, querying Mendeley’s global catalog, DOI lookups, and adding documents to your library. Tools available include mendeley_search_library, mendeley_get_document, mendeley_list_documents, mendeley_list_folders, mendeley_search_catalog, mendeley_get_by_doi, and mendeley_add_document. To integrate with Claude Desktop, configure your MCP server in Claude’s MCP settings by pointing to the mendeley-mcp entry, and restart Claude to enable the tools in your workspace. You can also run the server locally and connect any MCP client that supports the protocol by supplying the server URL and credentials.
Each tool maps to a specific API operation: search_library lets you query your local Mendeley library by title, author, abstract, or notes; get_document fetches complete metadata for a document; list_documents and list_folders help you navigate your collection; search_catalog queries Mendeley’s global repository; get_by_doi locates papers by DOI; and add_document allows creating new entries in your library. Authentication is handled via Mendeley OAuth credentials, stored securely in your system keyring or provided through environment variables in your MCP config.
How to install
Prerequisites:
- Python 3.10+ and pip
- A Mendeley account with API access (Client ID, Client Secret) and a registered Mendeley API app
Installation steps:
- Install the MCP server package from PyPI:
pip install mendeley-mcp
- Authenticate and set up credentials (follow the in-app wizard or use environment variables as described in the README):
- Run authentication:
mendeley-auth login
- Or export credentials via environment variables:
export MENDELEY_CLIENT_ID="your-client-id"
export MENDELEY_CLIENT_SECRET="your-client-secret"
export MENDELEY_REFRESH_TOKEN="your-refresh-token"
- Start the MCP server locally (example):
mendeley-mcp
-
If you prefer using uv or a container, follow the uv or Docker commands from the README to run the server in those environments.
-
Optional: test with MCP Inspector to validate your server is discoverable by MCP clients:
npm install -g @modelcontextprotocol/inspector
npx @modelcontextprotocol/inspector mendeley-mcp
Additional notes
Environment variables help avoid storing credentials in Claude or other clients. Common variables include MENDELEY_CLIENT_ID, MENDELEY_CLIENT_SECRET, and MENDELEY_REFRESH_TOKEN or MENDELEY_ACCESS_TOKEN. If you run into credential issues, use the mendeley-auth commands to login or check the config shown by mendeley-auth show-env. When using Docker or uv, ensure the environment variables are correctly passed to the container or process. If Claude Desktop does not show the Mendeley tools, verify that the MCP config file path is correct for your OS and that the JSON configuration is valid (no trailing commas). The Mendeley API enforces rate limits per user; the server can refresh tokens automatically, but if you hit limits, pause and retry or spread requests over time.
Related MCP Servers
mcp-agent
Build effective agents using Model Context Protocol and simple workflow patterns
nerve
The Simple Agent Development Kit.
python-utcp
Official python implementation of UTCP. UTCP is an open standard that lets AI agents call any API directly, without extra middleware.
prometheus
A Model Context Protocol (MCP) server that enables AI agents and LLMs to query and analyze Prometheus metrics through standardized interfaces.
mcp -odoo
A Model Context Protocol (MCP) server that enables AI assistants to securely interact with Odoo ERP systems through standardized resources and tools for data retrieval and manipulation.
octave
OCTAVE protocol - structured AI communication with 3-20x token reduction. MCP server with lenient-to-canonical pipeline and schema validation.