GeneBe MCP for AI assistants

GeneBe MCP lets an AI assistant look up a genetic variant in GeneBe and answer from data, instead of guessing. It is a hosted, read-only Model Context Protocol (MCP) server: you add one URL to your assistant, and it can annotate and classify variants for you.

https://mcp.genebe.net/v1/mcp

Research use only

GeneBe MCP is not a medical device and its answers are not a diagnosis. Classifications must be reviewed by a qualified specialist before any clinical use.

What you can ask

• "What is chr6-160585140-T-G? Explain why it is classified that way."
• "Is NM_005577.4:c.4195A>C benign?"
• "Convert rs41272110 to hg38 and list its consequences on each transcript."
• "Show the MANE Select transcript and ClinGen gene–disease validity for TP53."
• "Lift chr17:7674220 from hg38 to hg19."

For each variant the assistant gets GeneBe's transparent, points-based ACMG classification with the reasoning behind every criterion, together with population frequencies (gnomAD), in-silico predictor scores, ClinVar, per-transcript consequences and links back to GeneBe.

Connect your assistant

No account and no API key are needed. Choose your client.

Claude Code

claude mcp add --transport http genebe https://mcp.genebe.net/v1/mcp

Claude (web and desktop)

Open Settings → Connectors, choose Add custom connector, and enter https://mcp.genebe.net/v1/mcp.

ChatGPT

Open chatgpt.com/plugins, click Add (+) → Add custom MCP, and enter https://mcp.genebe.net/v1/mcp.

Cursor

Add to ~/.cursor/mcp.json:

{ "mcpServers": { "genebe": { "url": "https://mcp.genebe.net/v1/mcp" } } }

VS Code

Add to .vscode/mcp.json:

{ "servers": { "genebe": { "type": "http", "url": "https://mcp.genebe.net/v1/mcp" } } }

Any other MCP client

Any client that supports remote MCP over Streamable HTTP can use the same URL. To try it from a terminal:

curl -s -X POST https://mcp.genebe.net/v1/mcp \
  -H 'Content-Type: application/json' -H 'Accept: application/json, text/event-stream' \
  -d '{"jsonrpc":"2.0","id":1,"method":"tools/call","params":{"name":"annotate_variant","arguments":{"variant":"chr6-160585140-T-G"}}}'

You can also inspect the server with the MCP Inspector: choose the Streamable HTTP transport and enter the URL above.

Tools

Tool	What it does
annotate_variant	Start here. Any notation (genomic coordinates, HGVS, rsID, SPDI) gives the ACMG verdict, points and criteria, frequencies, predictor scores, ClinVar, consequences and GeneBe links.
explain_acmg_decision_path	Explains, in plain text, why each classification criterion applied.
convert_variant	Converts HGVS, rsID or SPDI to hg38 coordinates.
get_variant_effects	Per-transcript consequences (HGVS c./p., impact).
get_gene	Gene identifiers, location, MANE Select, ClinGen gene–disease validity and transcripts.
liftover_coordinate	Moves positions, intervals or variants between hg19, hg38 and T2T.

The default genome build is hg38. See variant formats for the notations GeneBe understands. Long lists are capped, so one call stays small enough for an assistant to read comfortably.

Privacy and limits

• Read-only. No tool changes anything.
• Single variants only. Do not send patient data or VCF files; the server does not accept them.
• Nothing is stored by the MCP server, and variant input is not logged.
• Fair use. Requests are rate limited per IP address.

For file-based work, use the GeneBe web application, the API or GeneBe Desktop.

Roadmap

Today GeneBe MCP answers questions about single variants. We plan to extend it so that an assistant can also work with your own data. None of the following is available yet, and the order may change.

• Sign-in, so the assistant can act on your GeneBe account
• Viewing and editing your own samples
• Searching for a variant across your own set of samples
• Uploading samples
• Starting analyses, following their progress and reading the results

Open source and feedback

The server is open source (MIT): github.com/genebe-net/genebe-mcp. It is also listed in the official MCP Registry as net.genebe/genebe-mcp. Questions, ideas and bug reports are welcome as GitHub issues or through the contact page.