Openalex

What openalex MCP server does

The openalex MCP server exposes OpenAlex’s structured scholarly graph to MCP-compatible clients. OpenAlex contains more than 240 million scholarly works and links them to authors, institutions, venues, concepts, funders, and citations. The data source is free and public, so no OpenAlex account or API key is required.

Use it to find a paper by title, DOI, or OpenAlex identifier and retrieve its structured record. It also supports searches involving authors and institutions, concept-based discovery, and citation questions such as which works cite a paper or which sources it references. A citable work URI follows the pattern pipeworx://openalex/work/{work_id}.

How it works

The hosted connection is available at https://gateway.pipeworx.io/openalex/mcp. A full Pipeworx gateway connection is also available at https://gateway.pipeworx.io/mcp; both reach the same underlying data sources, but the scoped endpoint lists OpenAlex tools directly. Hosted pack endpoints also expose shared Pipeworx meta-tools, including ask_pipeworx, discover_tools, search_within, and remember/recall.

The local package communicates over MCP using stdin and stdout. Run it with npx -y @pipeworx/mcp-openalex to expose only the OpenAlex pack’s tools, without the gateway’s shared meta-tools. Without an MCP client, the hosted gateway also provides HTTP tool access, including the documented openalex_search_works endpoint.

Setup and configuration

For a hosted setup, add the endpoint to an MCP client:

{
  "mcpServers": {
    "openalex": {
      "url": "https://gateway.pipeworx.io/openalex/mcp"
    }
  }
}

For local stdio use:

{
  "mcpServers": {
    "openalex": {
      "command": "npx",
      "args": ["-y", "@pipeworx/mcp-openalex"]
    }
  }
}

OpenAlex does not require authentication. The gateway automatically sends a polite-pool email and User-Agent on requests. No environment variables are documented as required.

Tools and capabilities

The openalex MCP server is suited to:

• Searching and retrieving scholarly works
• Looking up authors and institutions
• Browsing concepts and subject classifications
• Following cited and citing works
• Filtering research by publication context or open-access status
• Asking the gateway to route natural-language questions with ask_pipeworx

OpenAlex entity identifiers use stable prefixes: W for works, A for authors, I for institutions, V for venues, and C for concepts. Cross-entity queries can connect a paper to its authors, affiliations, publication venue, and concepts.

Limitations and notes

Author disambiguation is imperfect, particularly for common names or researchers whose records span separate identifiers. Cross-check with ORCID when available. Concepts are arranged in a six-level hierarchy; broad top-level concepts may be less useful than more specific lower-level entries.

Citation counts come from OpenAlex’s own indexed corpus and may differ from Google Scholar or Web of Science. Newly published works can take weeks to appear, while citations to recent works generally take longer. OpenAlex and Semantic Scholar are separate datasets, so their coverage and metadata can diverge. Open-access labels such as gold, green, hybrid, bronze, and closed describe the availability status tracked by OpenAlex, not a guarantee that every version is freely accessible.