Protein MCP Server

Query experimental and predicted protein structures through one keyless MCP server that federates RCSB PDB, AlphaFold DB, 3D-Beacons, UniProt, InterPro, and Foldseek.

   
Type MCP server
Supplier cyanheads
Availability GA
Pricing Free / OSS (Apache-2.0)
Capabilities Read-only — searches, fetches, aligns, and annotates public structure/sequence data; no writes
Verified works · 2026-07-27 — npm 0.4.0 resolves; bin protein-mcp-server confirms the npx launch
Security cleared · 2026-07-27 — provenance matches cyanheads, Apache-2.0, maintained, no OSV advisories

How to install

  • Claude Code — direct MCP add (stdio via npx): 
    claude mcp add --transport stdio protein-mcp-server -- npx -y @cyanheads/protein-mcp-server@latest
  • Claude Code — public hosted HTTP instance: 
    claude mcp add --transport http protein https://protein.caseyjhand.com/mcp
  • Claude Desktop — stdio via npx in claude_desktop_config.json: 
    {
  "mcpServers": {
    "protein-mcp-server": {
      "type": "stdio",
      "command": "npx",
      "args": ["-y", "@cyanheads/protein-mcp-server@latest"],
      "env": {
        "MCP_TRANSPORT_TYPE": "stdio",
        "MCP_LOG_LEVEL": "info"
      }
    }
  }
}
  • Claude Desktop — to use the hosted HTTP instance instead (Desktop has no native HTTP transport), proxy it with mcp-remote: 
    {
  "mcpServers": {
    "protein": {
      "command": "npx",
      "args": ["-y", "mcp-remote", "https://protein.caseyjhand.com/mcp"]
    }
  }
}

npx fetches the package on first launch; no separate pip/npm install step is required. Claude Code/Desktop launches the stdio process itself — there is no long-running server to keep open in another terminal.

What it does

Exposes seven tools over public structure and annotation APIs:

  • protein_search_structures — search experimental (PDB) and predicted structures.
  • protein_get_structure — fetch metadata and coordinate files by ID.
  • protein_find_similar — find sequence homologs (RCSB mmseqs2) or fold homologs (Foldseek) from a sequence, PDB ID, or UniProt accession.
  • protein_track_ligands — ligand discovery and binding-site analysis.
  • protein_compare_structures — structurally align 2–10 structures via TM-align / jFATCAT.
  • protein_analyze_collection — profile a set of PDB entries with server-side facets.
  • protein_get_annotations — fetch UniProt features and InterPro domains.

Primary use cases: federated experimental-plus-predicted structure lookup, sequence/fold homolog search, multi-structure alignment, ligand and annotation retrieval.

Notes

Every upstream provider (RCSB PDB, AlphaFold DB, 3D-Beacons, UniProt, InterPro, Foldseek) is public and keyless, so the server runs out of the box with no configuration. Both stdio and Streamable HTTP transports are supported. It is a data-retrieval and comparison layer — pair it with PyMOL or the ChimeraX MCP Server to render or edit the structures it returns, and with AlphaFold for de-novo prediction. Overlaps but broadens the RCSB-only PDB MCP Server by federating predicted models and adding TM-align/Foldseek comparison.

Sources

Installed this tool?

Share feedback — install path, OS, errors, workarounds. The form opens with this tool pre-selected and a link back to this page.