Skip to content
Open
9 changes: 9 additions & 0 deletions assets/icon-sources.json
Original file line number Diff line number Diff line change
Expand Up @@ -2559,5 +2559,14 @@
"homepage": "https://www.liblib.tv",
"mimeType": "image/png",
"sha256": "ada9717d540928baec64105056b93e8e8bc6cfbff145f9f3a9eadb6dd9b90689"
},
{
"name": "nimbus-bci",
"icon": "nimbus-bci/icon.png",
"source": "https://github.com/nimbusbci.png",
"matchedBy": "official-github-organization-avatar",
"homepage": "https://docs.nimbusbci.com",
"mimeType": "image/png",
"sha256": "56c89966f424e292b01018adac66c72f65892d0ad97849345bef2ca9ce1aa565"
}
]
Binary file added assets/nimbus-bci/icon.png
Loading
Sorry, something went wrong. Reload?
Sorry, we cannot display this file.
Sorry, this file is invalid so it cannot be displayed.
9 changes: 9 additions & 0 deletions external_plugins/nimbus-bci/.mcp.json
Original file line number Diff line number Diff line change
@@ -0,0 +1,9 @@
{
"mcpServers": {
"nimbus-bci": {
"type": "http",
"url": "https://nimbus-mcp.fly.dev/mcp",
"timeoutMs": 120000
}
}
}
18 changes: 18 additions & 0 deletions external_plugins/nimbus-bci/.zcode-plugin/plugin.json
Original file line number Diff line number Diff line change
@@ -0,0 +1,18 @@
{
"name": "nimbus-bci",
"description": "Nimbus BCI remote MCP server for brain-computer interface (BCI) and EEG decoding research on public datasets: 63 tools and 5 prompts for MOABB/BNCI datasets, pipeline build and validation, hosted training, plots, BIDS export, and a public leaderboard. Sign-in uses OAuth handled by the server.",
"description_i18n": {
"en": "Nimbus BCI remote MCP server for brain-computer interface (BCI) and EEG decoding research on public datasets: 63 tools and 5 prompts for MOABB/BNCI datasets, pipeline build and validation, hosted training, plots, BIDS export, and a public leaderboard. Sign-in uses OAuth handled by the server.",
"zh-CN": "Nimbus BCI 远程 MCP 服务,面向基于公开数据集的脑机接口(BCI)与 EEG 解码研究:提供 63 个工具和 5 个提示词,覆盖 MOABB/BNCI 数据集、流水线构建与校验、托管训练、绘图、BIDS 导出以及公开排行榜。登录通过服务端处理的 OAuth 完成。"
},
"version": "0.14.20",
"author": {
"name": "Nimbus BCI",
"email": "hello@nimbusbci.com",
"url": "https://github.com/nimbusbci"
},
"homepage": "https://docs.nimbusbci.com",
"repository": "https://github.com/nimbusbci/nimbus-mcp",
"license": "Apache-2.0",
"keywords": ["bci", "eeg", "neuroscience", "moabb", "bids", "machine-learning", "mcp"]
}
38 changes: 38 additions & 0 deletions external_plugins/nimbus-bci/README.md
Original file line number Diff line number Diff line change
@@ -0,0 +1,38 @@
# nimbus-bci

[中文文档](./README_CN.md)

Connects ZCode to the hosted [Nimbus BCI](https://docs.nimbusbci.com) MCP server, an Apache-2.0 toolkit for brain-computer interface (BCI) and EEG decoding research on public datasets.

## What it provides

- **MCP server `nimbus-bci`** (remote, Streamable HTTP): `https://nimbus-mcp.fly.dev/mcp`, with 63 tools and 5 prompts (nimbus-mcp 0.14.20):
- browse and load public MOABB / BNCI datasets;
- build and validate decoding pipelines;
- hosted model training;
- plots of data and results;
- BIDS export;
- a public results leaderboard.
- **Skill `nimbus-bci`**: guides the agent through the dataset → pipeline → validate → train → plot workflow and requires confirmation before any public leaderboard submission.

## Authentication

The server handles sign-in with OAuth. On first use ZCode opens the Nimbus authorization page in your browser. The plugin contains no API keys or other secrets, and nothing needs to be configured.

## Network access and side effects

- All tool calls go over HTTPS to `nimbus-mcp.fly.dev`, operated by Nimbus BCI. Datasets are fetched and training runs on that hosted service, not on your machine.
- The plugin writes no local files and ships no hooks, commands or local executables.
- Leaderboard submissions are publicly visible.
- Scope: public offline datasets only. The plugin does not connect to EEG hardware or run live sessions on people.

## Links

- Source: <https://github.com/nimbusbci/nimbus-mcp> (Apache-2.0)
- PyPI: [`nimbus-mcp`](https://pypi.org/project/nimbus-mcp/) 0.14.20 (local stdio alternative)
- Docs: <https://docs.nimbusbci.com>
- Contact: hello@nimbusbci.com

## License

Apache License 2.0. Public datasets accessed through the server keep their own licenses.
38 changes: 38 additions & 0 deletions external_plugins/nimbus-bci/README_CN.md
Original file line number Diff line number Diff line change
@@ -0,0 +1,38 @@
# nimbus-bci

[English](./README.md)

将 ZCode 连接到托管的 [Nimbus BCI](https://docs.nimbusbci.com) MCP 服务。Nimbus BCI 是一个基于 Apache-2.0 协议的开源工具包,用于在公开数据集上进行脑机接口(BCI)与 EEG 解码研究。

## 功能

- **MCP 服务 `nimbus-bci`**(远程,Streamable HTTP):`https://nimbus-mcp.fly.dev/mcp`,提供 63 个工具和 5 个提示词(nimbus-mcp 0.14.20):
- 浏览并加载 MOABB / BNCI 公开数据集;
- 构建并校验解码流水线;
- 托管模型训练;
- 数据与结果绘图;
- BIDS 导出;
- 公开的结果排行榜。
- **Skill `nimbus-bci`**:引导智能体完成"数据集 → 流水线 → 校验 → 训练 → 绘图"的流程,并在任何公开排行榜提交前要求用户确认。

## 认证

服务端通过 OAuth 处理登录。首次使用时,ZCode 会在浏览器中打开 Nimbus 授权页面。插件中不包含 API 密钥或其他机密信息,也无需任何配置。

## 网络访问与副作用

- 所有工具调用都通过 HTTPS 发送到由 Nimbus BCI 运营的 `nimbus-mcp.fly.dev`。数据集获取与训练在该托管服务上进行,而不是在你的电脑上。
- 插件不写入本地文件,不包含 hooks、命令或本地可执行文件。
- 排行榜提交内容公开可见。
- 范围:仅限离线公开数据集。插件不连接 EEG 硬件,也不对人进行实时会话。

## 链接

- 源码:<https://github.com/nimbusbci/nimbus-mcp>(Apache-2.0)
- PyPI:[`nimbus-mcp`](https://pypi.org/project/nimbus-mcp/) 0.14.20(本地 stdio 替代方案)
- 文档:<https://docs.nimbusbci.com>
- 联系方式:hello@nimbusbci.com

## 许可证

Apache License 2.0。通过该服务访问的公开数据集遵循其各自的许可证。
18 changes: 18 additions & 0 deletions external_plugins/nimbus-bci/skills/nimbus-bci/SKILL.md
Original file line number Diff line number Diff line change
@@ -0,0 +1,18 @@
---
name: nimbus-bci
description: Use when the user wants to explore public BCI/EEG datasets (MOABB, BNCI), build or validate a decoding pipeline, train a model on the hosted service, plot results, export to BIDS, or compare results on the Nimbus BCI leaderboard through the nimbus-bci MCP server.
---

# Nimbus BCI

Answer in the user's language. Use the `nimbus-bci` MCP tools and prompts:

1. Start from public datasets (MOABB / BNCI). Inspect a dataset before building a pipeline.
2. Build a pipeline, then validate it before training. Custom code is validated by the server; do not ask the user to run untrusted code locally.
3. Run hosted training, then use the plot tools to summarize results.
4. Offer BIDS export when the user needs a standard data layout.
5. Leaderboard submissions are public. Always show the user what will be submitted and get explicit confirmation before calling a leaderboard submit tool.

Scope: public offline datasets only. This plugin does not connect to EEG hardware or run live sessions on people.

If a tool returns an authorization error, ask the user to re-run the OAuth sign-in for the `nimbus-bci` server.
41 changes: 41 additions & 0 deletions marketplace.json
Original file line number Diff line number Diff line change
Expand Up @@ -1443,6 +1443,47 @@
"mcp",
"verification"
]
},
{
"name": "nimbus-bci",
"source": "./external_plugins/nimbus-bci",
"displayName": "Nimbus BCI",
"displayName_i18n": {
"en": "Nimbus BCI",
"zh-CN": "Nimbus BCI"
},
"icon": "https://cdn-zcode.z.ai/zcode/official-plugin/assets/nimbus-bci/icon.png",
"description": "Nimbus BCI remote MCP server for brain-computer interface (BCI) and EEG decoding research on public datasets: 63 tools and 5 prompts for MOABB/BNCI datasets, pipeline build and validation, hosted training, plots, BIDS export, and a public leaderboard. Sign-in uses OAuth handled by the server.",
"description_i18n": {
"en": "Nimbus BCI remote MCP server for brain-computer interface (BCI) and EEG decoding research on public datasets: 63 tools and 5 prompts for MOABB/BNCI datasets, pipeline build and validation, hosted training, plots, BIDS export, and a public leaderboard. Sign-in uses OAuth handled by the server.",
"zh-CN": "Nimbus BCI 远程 MCP 服务,面向基于公开数据集的脑机接口(BCI)与 EEG 解码研究:提供 63 个工具和 5 个提示词,覆盖 MOABB/BNCI 数据集、流水线构建与校验、托管训练、绘图、BIDS 导出以及公开排行榜。登录通过服务端处理的 OAuth 完成。"
},
"version": "0.14.20",
"author": {
"name": "Nimbus BCI",
"email": "hello@nimbusbci.com",
"url": "https://github.com/nimbusbci"
},
"category": "developer-tools",
"keywords": [
"bci",
"eeg",
"neuroscience",
"moabb",
"bids",
"machine-learning",
"mcp"
],
"examplePrompts": [
"List the public motor-imagery EEG datasets I can use",
"Build and validate a CSP + LDA pipeline on BNCI2014-001, then train it and plot the accuracy"
],
"examplePrompts_i18n": {
"zh-CN": [
"列出我可以使用的运动想象 EEG 公开数据集",
"在 BNCI2014-001 上构建并校验一个 CSP + LDA 流水线,然后训练并绘制准确率"
]
}
}
]
}