BioContextAI Knowledgebase

BioContextAI Knowledgebase MCP Server provides a standardized, OpenAPI-compatible interface that lets AI systems query and integrate with biomedical resources. You can run it locally for development or access a remote MCP instance for testing, enabling agentic models to retrieve verified biomedical data and perform domain-specific tasks with consistent behavior.

How to use

You connect to the MCP server from your agent or client by pointing it at either the remote MCP endpoint or your local MCP instance. Use the server to query biomedical data sources, retrieve metadata, and perform domain-specific actions through the common Model Context Protocol interface. When you run locally, you avoid remote rate limits and gain consistent availability for development and testing. When you use the remote endpoint, you can prototype quickly without setting up infrastructure.

How to install

Prerequisites: you need a supported runtime and a way to start MCP servers locally. The following options are provided for BioContextAI Knowledgebase MCP.

1. Local setup with Claude Desktop. Edit your MCP config to add the server entry for BioContextAI Knowledgebase MCP. The example below shows the configuration you would place in your Claude Desktop local MCP settings.

{
  "mcpServers": {
    "biocontext_kb": {
      "command": "uvx", // On Mac, use "brew install uv"
      "args": [
        "biocontext_kb@latest"
      ],
      "env": {
        "UV_PYTHON": "3.12"
      }
    }
  }
}

2. Local setup with uv. Start the server using streamable HTTP and uvicorn.

export MCP_ENVIRONMENT=PRODUCTION && export PORT=8000 && uvx biocontext_kb

3. Local setup with IDEs. Configure your coding agent to point at the MCP server using your editor’s MCP settings. For example, in VS Code you would add a server entry under the appropriate MCP settings file with the command and arguments shown in the IDE guidance.

4. Docker. Build and run a container to host the MCP server locally.

# Build the docker container
docker build -t biocontext_kb:latest .
# Run the container, exposing the MCP endpoint
docker run -p 127.0.0.1:8000:8000 biocontext_kb:latest

Public deployments should use a reverse proxy and enforce minimal access. Consider DDOS protection, restricted user permissions, and rootless container execution.

Additional sections

OpenAPI MCP servers can be created from configuration to extend the available tools. You can add new services by editing the configuration file that defines your OpenAPI-based MCP endpoints.

Tools exposed by BioContextAI Knowledgebase MCP include access to antibody registries, literature databases, pathway and protein resources, and more. Each tool represents an external biomedical data source the MCP can query as part of a unified interface.

Security and licensing notes: data sources have their own licenses. Ensure you comply with each source’s terms when using data in research or publication.

Available tools

Antibody Registry - Gene id conversion

Convert gene identifiers using the Antibody Registry data.

bioRxiv/medRxiv search

Search recent preprints and access metadata.

Ensembl id conversion

Convert gene identifiers and query Ensembl data.

EuropePMC search

Literature search and metadata access from EuropePMC.

Google Scholar lookup

Academic publications and author search (local use only).

InterPro classification

Protein families, domains, and functional sites classification.

KEGG pathways and interactions

Explore pathways and drug–drug interactions (licensing restrictions may apply).

OpenTargets associations

Target–disease associations across datasets.

PanglaoDB markers

Single-cell RNA-seq cell type marker lookup.

PRIDE proteomics data

Access mass spectrometry proteomics data.

Protein Atlas expression

Protein expression data lookup.

Reactome pathways

Curated biological pathways and annotations.

STRING interactions

Protein–protein interaction networks.

AlphaFold structures

Tertiary protein structure predictions.