Better Mealie MCP
Provides complete access to Mealie's API for managing recipes, meal plans, shopping lists, households, and more via natural language from AI assistants.
README
<div align="center">
🍲 Better Mealie MCP
<p> <img src="https://img.shields.io/badge/tools-259-blue" alt="259 tools"> <a href="https://github.com/jlowin/fastmcp"><img src="https://img.shields.io/badge/Built%20with-FastMCP-purple" alt="Built with FastMCP"></a> <img src="https://img.shields.io/badge/python-3.14%2B-blue" alt="Python 3.14+"> <img src="https://img.shields.io/badge/Mealie-v3.20.1-brightgreen" alt="Mealie v3.20.1"> </p> <p> <a href="https://github.com/djwmarcx/better-mealie-mcp/actions/workflows/pages.yml"><img src="https://img.shields.io/github/actions/workflow/status/djwmarcx/better-mealie-mcp/pages.yml?branch=main&label=setup%20wizard" alt="Setup wizard deploy"></a> <a href="https://github.com/djwmarcx/better-mealie-mcp/actions/workflows/update-spec.yml"><img src="https://img.shields.io/github/actions/workflow/status/djwmarcx/better-mealie-mcp/update-spec.yml?label=spec%20sync" alt="Spec sync"></a> <a href="https://github.com/djwmarcx/better-mealie-mcp/releases"><img src="https://img.shields.io/github/v/release/djwmarcx/better-mealie-mcp?sort=semver&label=release" alt="Latest release"></a> <a href="https://github.com/djwmarcx/better-mealie-mcp/blob/main/.github/renovate.json5"><img src="https://img.shields.io/badge/Renovate-enabled-brightgreen?logo=renovatebot&logoColor=white" alt="Renovate enabled"></a> </p>
<em>An MCP server exposing <strong>every</strong> <a href="https://mealie.io">Mealie</a> API endpoint —<br> all 250+ operations, none excluded. Manage recipes, meal plans, shopping lists,<br> households and more from any AI assistant, in natural language.</em>
</div>
Built with FastMCP from_openapi: tools are
generated straight from Mealie's OpenAPI spec, so the server stays in sync with
Mealie and nothing is hand-maintained. See TOOLS.md for the full
tool list.
💬 What can you do with it?
| You say | What happens |
|---|---|
| "Add a chicken tikka masala recipe from this URL" | Scrapes and imports the recipe |
| "What can I cook with what's in my pantry?" | Searches recipes by your ingredients |
| "Plan my dinners for next week" | Creates meal-plan entries |
| "Build a shopping list for those meals" | Generates a consolidated shopping list |
| "Tag all my soups as 'winter'" | Bulk-updates recipe tags |
🧙 Setup Wizard
<h3 align="center">
</h3>
<p align="center"> <a href="https://djwmarcx.github.io/better-mealie-mcp/"> <img src="public/options-tree.png" alt="Setup options — Install, Client, Transport, Auth" width="900"> </a> </p>
🚀 Setup
Docker (GHCR image):
docker pull ghcr.io/djwmarcx/better-mealie-mcp
docker run -i --rm \
-e MEALIE_BASE_URL=http://host.docker.internal:9925 \
-e MEALIE_API_TOKEN=... \
ghcr.io/djwmarcx/better-mealie-mcp # stdio; add `--http 8000` for HTTP
Images are published on each release, tagged <mealie-version> and latest.
Inside a container, localhost is the container — point MEALIE_BASE_URL at
host.docker.internal (macOS/Windows) or your host's LAN IP (Linux).
From source:
git clone https://github.com/djwmarcx/better-mealie-mcp
cd better-mealie-mcp
uv sync # install deps
cp .env.example .env # then edit .env with your Mealie URL + token
Auth (set in .env or the environment):
| Var | Meaning |
|---|---|
MEALIE_BASE_URL |
Mealie base URL (default http://localhost:9925) |
MEALIE_API_TOKEN |
Long-lived API token (preferred) — Mealie → Profile → Manage API Tokens |
MEALIE_USERNAME / MEALIE_PASSWORD |
Alternative: logs in at startup to fetch a token |
MEALIE_TIMEOUT |
Per-request timeout, seconds (default 60) |
MEALIE_VERIFY_SSL |
Verify TLS cert; false to accept self-signed (default true) |
MCP_SERVER_NAME |
MCP name advertised to clients (default Mealie) |
▶️ Run
uv run better-mealie-mcp # stdio transport (for MCP clients)
uv run better-mealie-mcp --http 8000 # streamable-http on 127.0.0.1:8000
uv run server.py # same server, back-compat entry
fastmcp run fastmcp.json # via FastMCP project config (stdio)
fastmcp run fastmcp-http.json # via FastMCP project config (http)
In --http mode the bind address comes from MCP_HOST (default 127.0.0.1;
the Docker image sets 0.0.0.0 so -p port mapping works).
🧪 Test against a local Mealie (Docker)
docker run -d --name mealie -p 9925:9000 \
-e ALLOW_SIGNUP=true -e BASE_URL=http://localhost:9925 -e TZ=UTC \
ghcr.io/mealie-recipes/mealie:latest
Default admin login: changeme@example.com / MyPassword.
📝 Notes
- Exposing every endpoint is a lot of tools — a lot of idle context. Most clients handle it, but if yours caps tool counts or you want a leaner context, use FastMCP's tool-search or filter by tag — ask and it can be wired in.
Versioning
This MCP's version mirrors the Mealie version its spec targets — MCP
3.20.1 ⇒ Mealie v3.20.1. The server advertises it to clients, and
VERSIONS.md maps every release to its Mealie version and date.
openapi.json is a vendored copy of Mealie's spec. The
update-spec workflow runs daily and
auto-tracks the latest stable Mealie release (mealie:latest): it boots that
image, reads its real version from /api/app/about
(MEALIE_VERSION), pulls /openapi.json, regenerates
TOOLS.md + counts, and — only when the spec actually changed —
bumps the version and opens a pull request (main is protected, so every
change lands via PR). When that PR merges,
release-on-spec cuts a
release (spec attached,
notes listing added/removed tools). Volatile server-clock defaults are stripped
so an unchanged run is a true no-op.
To freeze on one release instead of tracking latest, set MEALIE_TAG_DEFAULT in
the workflow to a specific tag (e.g. v3.20.1), or run it manually with a
mealie_tag input (latest, nightly, or any tag).
- A few endpoints (
list_auth_oauth*) return 500 unless OIDC is configured on the Mealie side — that's Mealie behavior, not the server.
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。