Seqera MCP Server

What is Seqera MCP Server?

Seqera MCP Server is Seqera's official hosted endpoint (https://mcp.seqera.io/mcp) that lets AI assistants launch and monitor Nextflow pipelines on Seqera Platform, build Wave containers, and search nf-core modules and public sequencing data.

Seqera is the company behind Nextflow, the workflow manager widely used in bioinformatics, and Seqera Platform, its service for running and monitoring pipelines across cloud and HPC compute. The MCP server brings that ecosystem into chat and coding assistants such as Claude, Cursor and VS Code, so researchers and bioinformatics engineers can manage runs, find modules and pull reference data in plain language.

How to use Seqera MCP Server?

You need a Seqera Platform account. There are two ways to authenticate:

• OAuth 2.1 (recommended): your browser opens automatically to sign in through Seqera Platform when you first connect.
• Personal access token: for clients without OAuth support, create a Seqera Platform access token and send it as Authorization: Bearer <YOUR_ACCESS_TOKEN>.

Claude Code

From Seqera's docs:

claude mcp add --scope=user --transport=http seqera https://mcp.seqera.io/mcp

Then run /mcp in a session and complete the Seqera sign-in in your browser.

Claude Desktop and claude.ai

Open Settings, choose "Add connectors", then "Add custom connector", enter https://mcp.seqera.io/mcp and select OAuth as the authentication method. Sign-in happens in the browser.

Cursor

Create or edit ~/.cursor/mcp.json:

{
  "mcpServers": {
    "seqera": {
      "url": "https://mcp.seqera.io/mcp"
    }
  }
}

Restart Cursor; your browser opens for authentication on first use.

VS Code

Add this to your user mcp.json (on macOS ~/Library/Application Support/Code/User/mcp.json, on Windows %APPDATA%\Code\User\mcp.json):

{
  "servers": {
    "seqera": {
      "url": "https://mcp.seqera.io/mcp",
      "type": "http"
    }
  }
}

Codex and Windsurf

Seqera documents both. For Codex, enable rmcp_client = true under [features] in ~/.codex/config.toml, then run codex mcp add seqera --url https://mcp.seqera.io/mcp and codex mcp login seqera. Windsurf uses serverUrl in ~/.codeium/windsurf/mcp_config.json.

Token instead of OAuth

{
  "mcpServers": {
    "seqera": {
      "url": "https://mcp.seqera.io/mcp",
      "headers": {
        "Authorization": "Bearer <YOUR_ACCESS_TOKEN>"
      }
    }
  }
}

Key features of Seqera MCP Server

• Official server from Seqera, the team behind Nextflow.
• Seqera Platform integration to launch, monitor and manage Nextflow pipelines.
• Wave container service to create containerized environments from conda or pip packages.
• Search and run more than 1,000 standardized nf-core bioinformatics modules.
• Search and retrieve public sequencing data from NCBI SRA, EBI ENA and GEO.
• OAuth 2.1 sign-in through Seqera Platform, with a personal access token fallback.
• Documented setup for Claude Code, Claude Desktop, Cursor, VS Code, Codex and Windsurf.

Use cases of Seqera MCP Server

• Run management: "Launch the nf-core/rnaseq pipeline in my workspace, then check its status."
• Monitoring: "Which of my Seqera Platform runs failed this week, and at which step?"
• Environment setup: "Build a Wave container with samtools and bcftools from conda."
• Pipeline design: "Find nf-core modules for variant calling and recommend which to use for short-read data."
• Data discovery: "Search SRA for public RNA-seq runs of human liver tissue and list the accessions."
• Writing Nextflow code in an editor with an assistant that can look up modules and check runs directly.

Troubleshooting Seqera MCP Server

• Client does not open a browser for sign-in: your client may not support MCP OAuth. Use a Seqera Platform personal access token in an Authorization: Bearer header instead.
• Cursor shows the server but no tools: restart Cursor after editing mcp.json, then complete the browser authentication.
• Windsurf ignores the server: Windsurf expects serverUrl, not url.
• Codex cannot add the server: enable rmcp_client = true in ~/.codex/config.toml first, then run codex mcp login seqera.