Pangaea

What the Pangaea MCP server does

The Pangaea MCP server exposes PANGAEA metadata for earth and environmental science datasets. It supports full-text search, direct lookup by numeric PANGAEA identifier, DOI-based retrieval, recent-publication queries, and facet aggregations. The underlying source is PANGAEA’s Elasticsearch metadata endpoint at https://ws.pangaea.de/es/pangaea/panmd.

Search accepts optional query, result-count, and offset values. Dataset lookup returns metadata for one PANGAEA ID, while DOI lookup accepts values such as 10.1594/PANGAEA.999999. Recent-dataset queries can be constrained by result count and age in days. Facets can group or aggregate values for fields including author, parameter, and project.

How it works

The Pangaea MCP server can run as a hosted MCP endpoint or as a local package. The hosted option is exposed through Pipeworx at https://gateway.pipeworx.io/pangaea/mcp; the full gateway is also available at https://gateway.pipeworx.io/mcp. Both URLs use the same gateway and data-source catalog, but the pack endpoint lists PANGAEA tools directly.

A hosted pack connection also includes shared Pipeworx meta-tools, such as ask_pipeworx, discover_tools, search_within, and memory-related tools. These additional tools allow questions outside the PANGAEA pack to be routed across the wider catalog. The exact tool scope is reported by the connection’s initialize response.

The local package communicates over MCP using standard stdin and stdout messages. In that mode, only the PANGAEA pack tools are exposed, with no shared gateway meta-tools or cross-catalog routing.

Setup and configuration

For a hosted connection, add the gateway URL to an MCP client configuration:

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

The README identifies Claude Desktop, Cursor, and Windsurf as compatible clients. Initial hosted calls do not require an account. Tools can also be called over HTTP; for example, send a JSON POST request to https://gateway.pipeworx.io/v1/tools/pangaea_search. Tool details are available through the corresponding /v1/tools/ endpoint.

For local use, configure the package with npx:

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

Running npx -y @pipeworx/mcp-pangaea directly can verify that the local server starts. The standalone mode does not require a Pipeworx account or a gateway round trip.

Tools and capabilities

The Pangaea MCP server provides these pack-specific tools:

• search(q?, size?, from?) for full-text dataset searches.
• dataset(id) for metadata lookup by numeric PANGAEA ID.
• dataset_by_doi(doi) for metadata lookup by DOI.
• recent(size?, days?) for recently published datasets.
• facets(field, q?, size?) for aggregations on metadata fields.

The hosted gateway may list roughly 30 additional shared tools alongside these pack tools. That larger list reflects gateway functionality rather than extra PANGAEA-specific operations. The package is licensed under MIT.