fireboard-mcp
MCP server for interacting with Fireboard BBQ temperature monitoring. Enables querying devices, live probe temperatures, Drive fan controller status, and historical cook sessions.
README
Fireboard MCP
MCP server for the Fireboard BBQ temperature monitoring API. Exposes tools for querying devices, live probe temperatures, Drive fan controller status and control, and historical cook sessions from any MCP-compatible AI assistant.
Live endpoint: https://fireboard-mcp.up.railway.app/mcp
Authentication
This server uses OAuth 2.0 with PKCE and Dynamic Client Registration — the standard required by MCP clients such as ChatGPT and Claude.
Fireboard's own API has no OAuth support; it only issues tokens via a username/password exchange. This server wraps that exchange behind a proper OAuth 2.0 flow: when you connect your AI assistant it opens a login page hosted by the MCP server, you enter your Fireboard credentials, and the server exchanges them with the Fireboard API for an access token. Your credentials are forwarded directly and are never stored — only the resulting API token is held in memory to authenticate tool calls on your behalf.
Note: The token is stored in memory only. If the server restarts (e.g. after a deploy), you will need to re-authenticate.
Connecting your AI assistant
Visit https://fireboard-mcp.up.railway.app for setup instructions for ChatGPT, Claude, Claude Code, and Gemini.
The MCP endpoint is: https://fireboard-mcp.up.railway.app/mcp
Tools
Rate limit: 17 calls per 5-minute window. See Fireboard API docs.
| Tool | What it does | API calls |
|---|---|---|
list_devices |
All Fireboard devices on the account | Cached (2 min TTL); 1 call on cache miss |
get_realtime_temps |
Current probe readings for all devices or a named device | Cached (2 min TTL); 1 call on cache miss |
get_drive_status |
Real-time Fireboard Drive fan %, setpoint, and control mode | 1 |
set_drive_setpoint |
Set target temperature and/or control channel (auto mode) | 1–2 |
set_drive_speed |
Set fan to a fixed speed (manual mode) | 1 |
set_drive_off |
Turn the Drive fan off | 1 |
list_sessions |
Recent cook sessions | 1 |
get_session_detail |
Session metadata and cook notes | 1 |
get_session_chart |
Full temperature time-series | 1 |
get_all_session_data |
Metadata, notes, and time-series in one call | 2 |
Running locally
npm install
npm run dev
Server starts at http://localhost:3000. MCP endpoint: http://localhost:3000/mcp.
Environment variables
| Variable | Required | Default | Description |
|---|---|---|---|
PUBLIC_DOMAIN |
In production | localhost:3000 |
Hostname (no protocol) used to build OAuth redirect URLs and the MCP endpoint. Must be set correctly in production — wrong value breaks OAuth. |
PORT |
No | 3000 |
Port the HTTP server listens on. |
NODE_ENV |
No | — | Set to production to enable JSON structured logging. |
FIREBOARD_API_BASE |
No | https://fireboard.io/api/v1 |
Override the Fireboard API base URL (useful for testing). |
FIREBOARD_CACHE_TTL_MS |
No | 120000 |
Device list cache TTL in milliseconds. |
推荐服务器
Baidu Map
百度地图核心API现已全面兼容MCP协议,是国内首家兼容MCP协议的地图服务商。
Playwright MCP Server
一个模型上下文协议服务器,它使大型语言模型能够通过结构化的可访问性快照与网页进行交互,而无需视觉模型或屏幕截图。
Magic Component Platform (MCP)
一个由人工智能驱动的工具,可以从自然语言描述生成现代化的用户界面组件,并与流行的集成开发环境(IDE)集成,从而简化用户界面开发流程。
Audiense Insights MCP Server
通过模型上下文协议启用与 Audiense Insights 账户的交互,从而促进营销洞察和受众数据的提取和分析,包括人口统计信息、行为和影响者互动。
VeyraX
一个单一的 MCP 工具,连接你所有喜爱的工具:Gmail、日历以及其他 40 多个工具。
graphlit-mcp-server
模型上下文协议 (MCP) 服务器实现了 MCP 客户端与 Graphlit 服务之间的集成。 除了网络爬取之外,还可以将任何内容(从 Slack 到 Gmail 再到播客订阅源)导入到 Graphlit 项目中,然后从 MCP 客户端检索相关内容。
Kagi MCP Server
一个 MCP 服务器,集成了 Kagi 搜索功能和 Claude AI,使 Claude 能够在回答需要最新信息的问题时执行实时网络搜索。
e2b-mcp-server
使用 MCP 通过 e2b 运行代码。
Neon MCP Server
用于与 Neon 管理 API 和数据库交互的 MCP 服务器
Exa MCP Server
模型上下文协议(MCP)服务器允许像 Claude 这样的 AI 助手使用 Exa AI 搜索 API 进行网络搜索。这种设置允许 AI 模型以安全和受控的方式获取实时的网络信息。