# Neuro MCP

> Neuro MCP is an open-source MCP server for health and science. MCP server giving AI agents 54 tools for EEG processing, EHR records and visualization. It runs locally on your machine and is started by your AI client.

- **Kind:** MCP server
- **Pricing:** Open source
- **Platforms:** CLI, self-hosted
- **Website:** https://aimplifier.github.io/neuro-mcp/
- **Repository:** https://github.com/aimplifier/neuro-mcp
- **MCP Registry name:** io.github.AImplifier/neuro-mcp
- **Categories:** health and science
- **Listing updated:** 2026-11-15

## About

Neuro-mcp is built for agents that help clinicians and researchers work with EEG data. A FastMCP server named neuro-analysis exposes 54 tools across five areas, so one agent can move from loading a recording to filtering, epoching and spectral analysis without leaving the same interface.

Processing tools work on an in-memory state keyed by a session id: load a recording once, then chain filter, ICA, epoch and spectral calls against it. Source imaging tools run on MNE-Python. A separate store keeps datasets, subjects, EHR records and annotations in SQLite by default or Postgres for production, with recordings laid out in BIDS on disk.

Clinical records are versioned and never overwritten or hard-deleted. An amendment adds a new version and keeps the prior one, a void marks a record entered-in-error, and every change is written to an audit log with actor, action and before and after values. Mutating tools take an explicit actor; the docs say auth and role enforcement are planned for v0.2. The NeuroII web visualization link is optional and returns a structured not-configured response when no endpoint is set.

## Features

- **EEG processing chain:** Load, filter, resample, re-reference, run ICA, epoch and compute PSD, ERP and time-frequency results in one session.
- **Source imaging:** Tools from fetching a template head to extracting label time courses support ESI workflows.
- **Versioned EHR records:** Amending creates a new version and voiding sets entered-in-error, so history is preserved.
- **Audit log:** Each mutation is recorded with actor, action and before and after values.
- **Dataset and BIDS store:** Register and query datasets and import recordings into a BIDS tree.
- **Interactive visualizations:** Produce standalone Plotly HTML views such as stacked multichannel EEG with annotation markers.

## Good for

- Review an EEG recording and add annotations through an agent
- Correct an EHR value while keeping the original and its author
- Run a preprocessing and spectral analysis pipeline on a dataset
- Run source imaging on a recording
- Push a recording to a NeuroII visualization session

## Install

### Claude Code (Terminal)

```sh
claude mcp add neuro-mcp -- uvx neuro-mcp
```

### Claude Desktop (claude_desktop_config.json)

```json
{
  "mcpServers": {
    "neuro-mcp": {
      "command": "uvx",
      "args": [
        "neuro-mcp"
      ]
    }
  }
}
```

### Cursor (~/.cursor/mcp.json)

```json
{
  "mcpServers": {
    "neuro-mcp": {
      "command": "uvx",
      "args": [
        "neuro-mcp"
      ]
    }
  }
}
```

### VS Code (.vscode/mcp.json)

```json
{
  "servers": {
    "neuro-mcp": {
      "type": "stdio",
      "command": "uvx",
      "args": [
        "neuro-mcp"
      ]
    }
  }
}
```

### Windsurf (~/.codeium/windsurf/mcp_config.json)

```json
{
  "mcpServers": {
    "neuro-mcp": {
      "command": "uvx",
      "args": [
        "neuro-mcp"
      ]
    }
  }
}
```

### Other clients (your client's MCP settings (most use this shape))

```json
{
  "mcpServers": {
    "neuro-mcp": {
      "command": "uvx",
      "args": [
        "neuro-mcp"
      ]
    }
  }
}
```

## Tools

- `load_neuro`: Load a neuro recording into a session
- `filter_neuro`: Filter a recording
- `resample_neuro`: Resample a recording
- `set_montage`: Set the electrode montage
- `set_reference`: Set the reference
- `detect_bad_channels`: Detect bad channels
- `run_ica`: Run ICA
- `apply_ica`: Apply ICA
- `find_events`: Find events
- `epoch_neuro`: Epoch a recording
- `compute_psd`: Compute power spectral density
- `compute_erp`: Compute event-related potentials
- `time_frequency`: Time-frequency analysis
- `register_subject`: Register a subject
- `get_subject`: Get a subject
- `add_ehr_record`: Add an EHR record
- `amend_ehr_record`: Amend an EHR record as a new version
- `get_ehr_history`: Get EHR history
- `void_ehr_record`: Mark an EHR record entered-in-error
- `import_recording`: Import a recording to BIDS
- `register_dataset`: Register a dataset
- `query_datasets`: Query datasets
- `list_recordings`: List recordings
- `add_annotation`: Add an annotation
- `update_annotation`: Update an annotation as a new version
- `list_annotations`: List annotations
- `void_annotation`: Void an annotation
- `get_audit_log`: Read the audit log
- `neuroii_push_recording`: Push a recording to NeuroII
- `neuroii_create_viz_session`: Create a NeuroII visualization session
- `neuroii_pull_annotations`: Pull annotations from NeuroII
- `visualize_timeseries`: Interactive stacked multichannel EEG view

## FAQ

### How do I install it?

It is published on PyPI as neuro-mcp (version 0.1.4 in the registry) and runs over stdio.

### Which database does it use?

SQLite by default, and Postgres for production, plus a BIDS recording tree on disk.

### Can records be deleted?

No. Records are versioned and voiding only marks them entered-in-error; history stays.

### Is access control enforced?

Not yet. The docs say auth and RBAC enforcement is planned for v0.2.

### Is Neuro MCP free?

Yes. Neuro MCP is open source, so you can use and self-host it for free.

### How do I add Neuro MCP to Claude Code?

Run this command in your terminal: claude mcp add neuro-mcp -- uvx neuro-mcp

### What are the best alternatives to Neuro MCP?

The closest alternatives to Neuro MCP are Tooluniverse, Oya Browser, Later Stages, MedXpert 标准检索连接器 and MedXpert 全球法规连接器 (reg-connector).

## Alternatives

- [Tooluniverse](https://ai-powered.com/mcp/tooluniverse): Tooluniverse is an open-source MCP server for health and science.
- [Oya Browser](https://ai-powered.com/mcp/oya-browser): Oya Browser is a freemium MCP server for browser automation.
- [Later Stages](https://ai-powered.com/mcp/later-stages): Later Stages is a free MCP server for health and science.
- [MedXpert 标准检索连接器](https://ai-powered.com/mcp/medxpert-2): MedXpert 标准检索连接器 is a free MCP server for legal and government.
- [MedXpert 全球法规连接器 (reg-connector)](https://ai-powered.com/mcp/medxpert-reg-connector): MedXpert 全球法规连接器 (reg-connector) is a free MCP server for legal and government.
- [Splyt](https://ai-powered.com/mcp/splyt): Splyt is a free MCP server for health and science.
- [Health Agent](https://ai-powered.com/mcp/health-agent): Health Agent is an MCP server for health and science.
- [Pausila](https://ai-powered.com/mcp/pausila): Pausila is a free MCP server for local services.

---
Source: https://ai-powered.com/mcp/neuro-mcp
