MNE MCP
Let your coding agent run MNE-Python EEG/MEG and related neurophysiology workflows through MCP instead of juggling notebooks and shell one-offs.
Overview
MNE-MCP is a Build-phase MCP server that connects coding agents to MNE-Python for EEG, MEG, sEEG, ECoG, and fNIRS analysis over stdio.
What is this MCP server?
- MNE-Python neurophysiology exposed through Model Context Protocol stdio
- Modalities called out in catalog: EEG, MEG, sEEG, ECoG, fNIRS
- PyPI package mne-mcp v0.2.2 with uvx runtimeHint and serve subcommand
- Optional ICA extra via --from mne-mcp[ica] runtime argument
- stdio transport flag --transport stdio for agent clients
- MCP server version 0.2.2
- PyPI identifier mne-mcp with uvx runtimeHint
- Documented modalities: EEG, MEG, sEEG, ECoG, fNIRS
Community signal: 1 GitHub stars.
What problem does it solve?
Neurophysiology pipelines are painful to drive from an AI coding agent when every preprocessing step lives in opaque notebooks outside MCP.
Who is it for?
Indie researchers or health-tech builders who already use MNE-Python and want agent-driven, repeatable analysis via uvx and stdio MCP.
Skip if: General SaaS analytics, builders with no neuro data, or teams needing a governed clinical UI without Python expertise.
What do I get? / Deliverables
After registering MNE-MCP, your agent can invoke MNE-backed analysis tools through MCP while you stay in one harness-led workflow.
- Agent-callable MNE-Python analysis operations via MCP tools
- Repeatable neuro pipelines without leaving the coding harness
- Optional ICA-enabled runs when using mne-mcp[ica] runtime argument
Recommended MCP Servers
Journey fit
Neurophysiology analysis tooling is integrated while you are building research or health-tech pipelines; Validate rarely needs a full MNE surface, so Build is the primary shelf. The server wraps MNE-Python as an external analysis engine—classic integrations work connecting an agent to a specialized Python stack via uvx/PyPI.
How it compares
Domain scientific-data MCP integration, not a generic SQL or spreadsheet connector.
Common Questions / FAQ
Who is MNE MCP for?
Developers and researchers building neurophysiology tooling who want Claude Code, Cursor, or similar agents to call MNE-Python through MCP.
When should I use MNE MCP?
During Build when you are wiring an agent into EEG/MEG/sEEG/ECoG/fNIRS preprocessing, visualization, or analysis scripts you would otherwise run manually in Python.
How do I add MNE MCP to my agent?
Configure an MCP stdio server that runs uvx with mne-mcp serve --transport stdio (optionally --from mne-mcp[ica] for ICA extras), then restart your client and use the exposed tools on your local datasets.