← Back to directory
COMPARE UP TO 4 SERVERS

Compare MCP servers

Compare scores, permissions, risks, and fit in one decision-focused table.

DimensionArxiv MCP ServerSearch, download, and analyze arXiv papers through MCP.BioMCPOne binary. One grammar. Evidence from the biomedical sources you already trust.
FMRS64 / 100 · C75 / 100 · B
Reliability10 / 2010 / 20
Security and permissions13 / 2016 / 20
Maintenance11 / 2018 / 20
Documentation16 / 2017 / 20
Setup experience14 / 2014 / 20
Best for
  • Researchers who want AI-assisted arXiv literature discovery
  • Users who need local paper storage and full-text or semantic search
  • Research workflows able to review external paper content and manage local storage
  • Biomedical researchers
  • Clinicians
  • AI agents that need efficient access to biomedical literature and data analysis
Not for
  • Users who only want browser-based search without local downloads
  • Sensitive workflows that cannot accept paper text entering model context
  • Deployments requiring proven official maintenance or enterprise authentication
  • Users needing to write to upstream databases (read-only)
  • Users requiring direct access to closed datasets (e.g., COSMIC is not directly integrated)
  • Users expecting all advanced features without API keys
Required permissions
  • Access arXiv to search for and download papers
  • Read and write papers in the configured local storage directory
  • Access Semantic Scholar when using citation_graph
  • Listen on the configured local host and port in HTTP mode
  • Requires internet access to query public biomedical APIs
  • Optional API keys (NCBI, Semantic Scholar, OpenFDA, etc.) to improve rate limits or unlock features
  • Local read/write access for skills, cache, and local study data (e.g., `~/.local/share/biomcp`)
Risks and side effects
  • Paper content is untrusted external input and may contain prompt-injection instructions
  • Combining this server with filesystem, shell, or browser tools increases tool-chain attack risk
  • semantic_search, citation_graph, alerts, and related features are experimental and may behave unexpectedly
  • arXiv rate limits can cause search requests to fail temporarily
  • Upstream API rate limits may affect performance
  • Some data sources (e.g., KEGG) have licensing terms that restrict non-academic use
  • Features requiring API keys (e.g., OncoKB) may be incomplete without them
  • Local binary installations need regular updates for compatibility
Supported clientsClaude Desktop, CodexClaude Desktop, Claude Code, Codex, Other MCP stdio clients
Tools1111