ai.smithery/pinion05-supabase-mcp-lite
AI 与智能体by pinion05
提供相同功能,但仅消耗约 1/20 的 context window tokens,更节省上下文资源。
什么是 ai.smithery/pinion05-supabase-mcp-lite?
提供相同功能,但仅消耗约 1/20 的 context window tokens,更节省上下文资源。
README
Supabase MCP Lite
<img width="518" height="141" alt="image" src="https://github.com/user-attachments/assets/04193768-632c-4e32-93d2-291158c804b5" />Minimal Supabase MCP server - 70% less context usage than standard implementations.
Why Lite?
- 4 tools instead of 50+ - Only essential operations
- Minimal descriptions - No verbose explanations
- Simple parameters - No complex nested schemas
- Auto-truncated results - Max 100 rows per query
🔑 Personal Access Token Required
This MCP uses your Supabase Personal Access Token (starts with sbp_) to automatically fetch service role keys for any project you own.
How to get your Personal Access Token:
- Go to https://supabase.com/dashboard/account/tokens
- Click "Generate New Token"
- Give it a name (e.g., "MCP Access")
- Copy the token (starts with
sbp_) - Save it securely - you won't be able to see it again!
Setup
- Add to your MCP client configuration:
{
"supabase-lite": {
"command": "npx",
"args": ["@smithery/cli", "connect", "@pinion05/supabase-mcp-lite"],
"config": {
"accessToken": "sbp_xxxxxxxxxxxx" // Your Personal Access Token
}
}
}
Note: Project URL is required for each tool call. The service role key will be fetched automatically using your access token.
Tools (4)
All tools require projectUrl as the first parameter.
| Tool | Purpose | Parameters |
|---|---|---|
select | Get data | projectUrl, table, where?, limit? |
mutate | Change data | projectUrl, action, table, data?, where? |
storage | Files | projectUrl, action, bucket, path?, data? |
auth | Users | projectUrl, action, email?, password?, id? |
Examples
// Select tool
select(
projectUrl: "https://your-project.supabase.co",
table: "posts",
where: {status: "published"},
limit: 10
)
// Mutate tool
mutate(
projectUrl: "https://your-project.supabase.co",
action: "insert",
table: "todos",
data: {title: "New task"}
)
mutate(
projectUrl: "https://your-project.supabase.co",
action: "update",
table: "todos",
data: {done: true},
where: {id: 1}
)
mutate(
projectUrl: "https://your-project.supabase.co",
action: "delete",
table: "todos",
where: {id: 1}
)
// Storage tool
storage(
projectUrl: "https://your-project.supabase.co",
action: "upload",
bucket: "images",
path: "avatar.jpg",
data: "base64..."
)
storage(
projectUrl: "https://your-project.supabase.co",
action: "list",
bucket: "images"
)
// Auth tool
auth(
projectUrl: "https://your-project.supabase.co",
action: "list"
)
auth(
projectUrl: "https://your-project.supabase.co",
action: "create",
email: "user@example.com",
password: "secure123"
)
How it Works
- You provide your Personal Access Token (
sbp_xxx) - When you call a tool with a project URL, the MCP:
- Extracts the project ID from the URL
- Uses your access token to fetch the service role key via Supabase Management API
- Caches the key for future requests to the same project
- Creates a client with full admin access
Security Notes
- Personal Access Token gives access to ALL your Supabase projects
- Service role keys are fetched automatically and cached in memory
- Service role key bypasses Row Level Security (RLS)
- Keep your access token secure - never expose it client-side
- This tool is intended for server-side/admin use only
Features
- ✅ Works with any Supabase project you own
- ✅ Automatic service role key retrieval
- ✅ Key caching to minimize API calls
- ✅ Full database access (bypasses RLS)
- ✅ Support for multiple projects in one session
License
MIT
常见问题
ai.smithery/pinion05-supabase-mcp-lite 是什么?
提供相同功能,但仅消耗约 1/20 的 context window tokens,更节省上下文资源。
相关 Skills
Claude接口
by anthropics
面向接入 Claude API、Anthropic SDK 或 Agent SDK 的开发场景,自动识别项目语言并给出对应示例与默认配置,快速搭建 LLM 应用。
✎ 想把Claude能力接进应用或智能体,用claude-api上手快、兼容Anthropic与Agent SDK,集成路径清晰又省心
RAG架构师
by alirezarezvani
聚焦生产级RAG系统设计与优化,覆盖文档切块、检索链路、索引构建、召回评估等关键环节,适合搭建可扩展、高准确率的知识库问答与检索增强应用。
✎ 面向RAG落地,把知识库、向量检索和生成链路系统串联起来,做架构设计时更清晰,也更少踩坑。
计算机视觉
by alirezarezvani
聚焦目标检测、图像分割与视觉系统落地,覆盖 YOLO、DETR、Mask R-CNN、SAM 等方案,适合定制数据集训练、推理优化及 ONNX/TensorRT 部署。
✎ 把目标检测、图像分割到推理部署串成完整工程链路,主流框架与 YOLO、DETR、SAM 等方案都覆盖,落地视觉 AI 会省心很多。
相关 MCP Server
顺序思维
编辑精选by Anthropic
Sequential Thinking 是让 AI 通过动态思维链解决复杂问题的参考服务器。
✎ 这个服务器展示了如何让 Claude 像人类一样逐步推理,适合开发者学习 MCP 的思维链实现。但注意它只是个参考示例,别指望直接用在生产环境里。
知识图谱记忆
编辑精选by Anthropic
Memory 是一个基于本地知识图谱的持久化记忆系统,让 AI 记住长期上下文。
✎ 帮 AI 和智能体补上“记不住”的短板,用本地知识图谱沉淀长期上下文,连续对话更聪明,数据也更可控。
PraisonAI
编辑精选by mervinpraison
PraisonAI 是一个支持自反思和多 LLM 的低代码 AI 智能体框架。
✎ 如果你需要快速搭建一个能 24/7 运行的 AI 智能体团队来处理复杂任务(比如自动研究或代码生成),PraisonAI 的低代码设计和多平台集成(如 Telegram)让它上手极快。但作为非官方项目,它的生态成熟度可能不如 LangChain 等主流框架,适合愿意尝鲜的开发者。