Model Context Protocol moderate risk

Biomcp

BioMCP empowers AI agents with biomedical knowledge via structured access to clinical trials, literature, and genomic data, enhancing research and clinical decision support.

Connections & Capabilities

Connects To

GitHubDocker

Capabilities

read

Quickstart

Install

npx @modelcontextprotocol/inspector uv run --with biomcp-python biomcp run

Config

{
  "mcpServers": {
    "biomcp": {
      "command": "uv",
      "args": ["run", "--with", "biomcp-python", "biomcp", "run"]
    }
  }
}

Exposed MCP Tools (22)

safe

think

Breaks down complex biomedical queries into sequential steps.

Purely a planning tool with no external interaction.

safe

search

Searches across various biomedical databases using a unified query language or domain-specific filters.

Read-only access to indexed data; no data modification.

safe

fetch

Retrieves detailed information for a specific article, trial, or variant.

Retrieves existing data; no modification or creation.

safe

article_searcher

Searches PubMed/PubTator3 and preprint servers for biomedical articles.

Read-only access to article databases.

safe

article_getter

Fetches detailed information for a specific article using PMID or DOI.

Read-only access to article details.

safe

trial_searcher

Searches ClinicalTrials.gov or NCI CTS API for clinical trials.

Read-only access to trial databases.

safe

trial_getter

Fetches all details for a specific clinical trial from either source.

Read-only access to trial details.

safe

trial_protocol_getter

Fetches protocol information only for a specific clinical trial.

Read-only access to trial protocols.

safe

trial_references_getter

Fetches publications related to a specific clinical trial.

Read-only access to trial references.

safe

trial_outcomes_getter

Fetches outcome measures and results for a specific clinical trial.

Read-only access to trial outcomes.

safe

trial_locations_getter

Fetches site locations and contact information for a specific clinical trial.

Read-only access to trial locations.

safe

variant_searcher

Searches the MyVariant.info database for genetic variants.

Read-only access to variant data.

safe

variant_getter

Fetches comprehensive details for a specific genetic variant.

Read-only access to variant details.

safe

nci_organization_searcher

Searches NCI's organization database.

Read-only access to organization data.

safe

nci_organization_getter

Gets organization details by ID from NCI's database.

Read-only access to organization details.

safe

nci_intervention_searcher

Searches NCI's intervention database (drugs, devices, procedures).

Read-only access to intervention data.

safe

nci_intervention_getter

Gets intervention details by ID from NCI's database.

Read-only access to intervention details.

safe

nci_biomarker_searcher

Searches biomarkers used in trial eligibility criteria.

Read-only access to biomarker data.

safe

nci_disease_searcher

Searches NCI's controlled vocabulary of cancer conditions.

Read-only access to disease data.

safe

gene_getter

Gets real-time gene information from MyGene.info.

Read-only access to gene information.

safe

disease_getter

Gets disease definitions and synonyms from MyDisease.info.

Read-only access to disease definitions.

safe

drug_getter

Gets drug/chemical information from MyChem.info.

Read-only access to drug information.

Safety Assessment

BioMCP is generally safe for read-only access to biomedical data. Risks are primarily associated with data interpretation and reliance on external APIs. It's crucial to validate results and ensure responsible use of sensitive information.

Primarily read-only access to biomedical databases.
No direct execution of arbitrary code.
Uses well-established biomedical data sources.
Clear separation of search and fetch operations.

Access to sensitive biomedical data requires careful handling.
Potential for misuse of information if not properly interpreted.
Reliance on external APIs introduces dependency risks.
Improperly constructed queries could lead to unintended data retrieval.