SciAgent (mcp-server-bioscience) 🧬🔬

SciAgent (BioScience MCP Server) is a high-performance, security-focused Model Context Protocol (MCP) server that empowers AI Agents with cutting-edge biological, medical, and scientific capabilities.

Traditional AI Agents are limited to simple web searches. By connecting your Agent (Cursor, Claude Desktop, Aider, etc.) to SciAgent, you instantly grant it the ability to:

• Deep-mine academic literature (PubMed, EuropePMC).

• Analyze genetic variants and genomic data (Ensembl, dbSNP).

• Retrieve and analyze 3D protein structures (UniProt, AlphaFold).

• Execute complex, multi-step biomedical workflows (e.g., Literature -> Gene -> Protein 3D Structure).

🛡️ Built for AI Safety (Codex Security Ready)

When AI processes third-party biological data (such as parsing PDB files or reading unverified scientific texts), it is highly susceptible to Prompt Injection and Buffer Overflow/Malicious Script Execution vulnerabilities.

SciAgent is engineered with security first. It features a built-in Sanitizer Middleware that strictly validates API responses, limits token exposure, and strips out potentially malicious payloads before handing data back to the LLM.

Related MCP server: grant-mcp

🚀 Quick Start

1. Installation

This package requires uv or pip. We highly recommend using uv for speed.

# Clone the repository
git clone https://github.com/YOUR_GITHUB_USERNAME/mcp-server-bioscience.git
cd mcp-server-bioscience

# Install
uv pip install -e .

2. Connect to Your AI (e.g., Claude Desktop)

Add the following configuration to your claude_desktop_config.json:

{
  "mcpServers": {
    "bioscience": {
      "command": "mcp-bioscience"
    }
  }
}

🛠️ Included Agent Skills

1. search_literature: Query PubMed for biomedical papers.

2. get_protein_structure: Fetch AlphaFold predicted 3D structures and pLDDT confidence scores.

3. get_gene_info: Retrieve genomic coordinates and metadata from Ensembl.

4. analyze_disease_target (Workflow): Automatically discover disease targets from literature and fetch their protein structures in one go.

🤝 Contributing

We welcome contributions! Please see CONTRIBUTING.md for guidelines. This project aims to be the standard bioinformatics toolkit for the global Agent ecosystem.