DataClawe — Database as a Utility

com.dataclawe由 dataclawe.com 发布官方发布者v1.1.0更新于 Oct 24, 2026

AI-native database utility. Talk to your DB in plain language. No SQL. MySQL & PostgreSQL.

已验证SSE可网页运行数据库数据与分析

服务详情

标识符: com.dataclawe/dataclawe-mcp
传输: SSE
运行时: Web executable
信任: Verified
发布于: Sep 16, 2026

安全提示

MCP 服务会接收对话中的工具调用请求，并可能执行外部操作。启用前请先了解该服务能做什么，并只授予它所需的权限。

注册中心已验证此服务的命名空间属于 dataclawe.com，且该域名同时运行其端点。这可以确认发布者，但并不代表该服务的工具是安全的。

SourceWeft 的检查只涵盖条目是否完整、端点是否响应；不审查服务器的代码，也不审查其工具的行为。

在 README 中添加徽章

向读者展示该 MCP 服务已收录于 SourceWeft，徽章链接到本页。

[![Listed on SourceWeft](https://sourceweft.com/badge/mcp/com.dataclawe%2Fdataclawe-mcp.svg)](https://sourceweft.com/mcp/com.dataclawe%2Fdataclawe-mcp?utm_source=readme&utm_medium=badge)

概览

AI 生成的概览

远程 MCP 服务器，让助手用 DataClawe 的 JSON 命令语言而非 SQL 来查询和修改 MySQL 或 PostgreSQL 数据。

可以做什么: 读取数据：按条件筛选、排序和限制条数，进行查询、计数、存在性判断与聚合
查看结构：列出数据表、读取字段定义、获取记录数、更新时间和存储大小等表统计
写入数据：插入、更新记录，以及逻辑删除记录
管理数据表：创建、复制、重命名和逻辑删除数据表
查看历史：获取某条记录按时间排列的变更历史
适用场景: 适合让助手在不写 SQL 的情况下操作已有的 MySQL 或 PostgreSQL 数据库，例如日常查询、报表取数，或在旧系统上做简单的记录维护。
运行要求: 远程 SSE 端点 mcp.dataclawe.com；需 OAuth 认证，通过 Authorization 请求头以 Bearer 格式携带 DataClawe 访问密钥，密钥需在厂商网站获取。隐含需要 DataClawe 账号及已接入的 MySQL 或 PostgreSQL 数据库。

安装前请注意

访问密钥是机密信息，通过 Authorization 请求头传输。写入类操作（插入、更新、删除、建表、复制、重命名、删表）会改动数据；删除为逻辑标记而非物理清除。服务收费：一次性注册费，加上按月套餐及按记录、按会话的超额费用。数据会发送给第三方转换服务。

安装

在 SourceWeft 中

打开控制台中的 DataClawe — Database as a Utility，将其添加到工作区。
连接所需的凭据。它们会被私密存储，绝不会在公开页面上展示。
为需要使用其工具的对话启用该服务。

Web executable，通过 SSE。远程服务在工作区中配置后即可从网页运行时运行。

交给智能体安装

把这段提示词粘贴到 Claude Code、Codex、Cursor 等智能体，或 SourceWeft 对话中。智能体会读取该服务器的安装指南，先向你展示发布者、端点和登录方式，经你同意后再把它添加到你的客户端。密钥不会写进任何命令。

阅读 https://sourceweft.com/mcp/com.dataclawe%2Fdataclawe-mcp/install.md，并按其中说明把这个 MCP 服务器接入我的客户端。修改任何配置之前，先告诉我它的发布者、端点和登录方式，等我确认。不要把 API 密钥写进命令或我会提交的文件；问我密钥应从哪里获取。

查看智能体所依据的安装指南

其他 MCP 客户端

把它添加到你客户端的 mcpServers 配置中。客户端会在首次连接时完成 OAuth 登录。

{
  "mcpServers": {
    "dataclawe-database-as-a-utility": {
      "type": "sse",
      "url": "https://mcp.dataclawe.com/sse"
    }
  }
}

README

DCL — DataClawe Command Language

A JSON-based database command standard for AI agents, frontend engineers, and anyone who finds SQL too complex.

"If you understand MySQL and MSX-BASIC, you already understand DCL."

What is DCL?

DCL (DataClawe Command Language) is an open standard for communicating with databases using simple JSON.

It is designed for:

AI agents (Claude, GPT, Cursor, and others) to safely read and write structured data
Frontend engineers who need database access without writing SQL
MCP (Model Context Protocol) tool integration
Any application connecting to MySQL or PostgreSQL — legacy or new

DCL translates into native SQL internally. Your existing database does not change. Your existing application does not change.

Design Philosophy

DCL follows three rules:

1. If you can read it, you understand it. No cryptic operators. No framework-specific syntax. No : chaining.

2. MySQL words. JSON structure. Actions like SELECT, INSERT, UPDATE, DELETE are exactly what you expect. WHERE conditions are written the way MySQL engineers already write them.

3. MSX-BASIC level simplicity. If a condition like age >= 20 needs an explanation, the design has failed.

4. Simple by default. Powerful when needed. Standard covers most real-world needs. Advanced covers the rest. You never pay the complexity cost until you need it.

Quick Start

Fetch data

json

{  "dcl": "1.0",  "action": "SELECT",  "table": "users",  "columns": ["id", "name", "email"],  "where": [    "status = 'active'",    "age >= 20"  ],  "order": "created_at DESC",  "limit": 10,  "offset": 0}

{  "dcl": "1.0",  "action": "SELECT",  "table": "users",  "columns": ["id", "name", "email"],  "where": [    "status = 'active'",    "age >= 20"  ],  "order": "created_at DESC",  "limit": 10,  "offset": 0}

Insert a record

json

{  "dcl": "1.0",  "action": "INSERT",  "table": "users",  "data": {    "name": "kimura",    "email": "[email protected]",    "status": "active"  }}

{  "dcl": "1.0",  "action": "INSERT",  "table": "users",  "data": {    "name": "kimura",    "email": "[email protected]",    "status": "active"  }}

Update records

json

{  "dcl": "1.0",  "action": "UPDATE",  "table": "users",  "data": {    "status": "inactive"  },  "where": [    "id = 123"  ]}

{  "dcl": "1.0",  "action": "UPDATE",  "table": "users",  "data": {    "status": "inactive"  },  "where": [    "id = 123"  ]}

Delete a record (logical delete)

Actions

Action	Description	Read-only
`SELECT`	Fetch one or more records	✅
`COUNT`	Count matching records	✅
`EXISTS`	Check if a record exists	✅
`SCHEMA`	Get column definitions for a table	✅
`TABLES`	List all available tables	✅
`STATS`	Get record count, last updated, and storage size	✅
`TIMELINE`	Get chronological change history of a record	✅
`INSERT`	Create a new record	❌
`UPDATE`	Update existing records	❌
`DELETE`	Logical delete (status flag)	❌
`TABLE_CREATE`	Create a new table (empty declaration)	❌
`TABLE_COPY`	Copy a table	❌
`TABLE_RENAME`	Rename a table	❌
`TABLE_DROP`	Logical delete a table	❌

WHERE Conditions

WHERE is an array of condition strings. Multiple conditions are AND by default.

Basic comparisons

OR conditions

AND + OR combined

json

"where": {  "AND": [    "age >= 20",    {      "OR": [        "status = 'active'",        "status = 'pending'"      ]    }  ]}

"where": {  "AND": [    "age >= 20",    {      "OR": [        "status = 'active'",        "status = 'pending'"      ]    }  ]}

LIKE

IN

BETWEEN

NULL checks

Aggregation

json

{  "dcl": "1.0",  "action": "SELECT",  "table": "orders",  "columns": [    "SUM(amount) AS total",    "AVG(amount) AS average",    "COUNT(*) AS count"  ],  "where": [    "status = 'paid'"  ],  "group": "customer_id"}

{  "dcl": "1.0",  "action": "SELECT",  "table": "orders",  "columns": [    "SUM(amount) AS total",    "AVG(amount) AS average",    "COUNT(*) AS count"  ],  "where": [    "status = 'paid'"  ],  "group": "customer_id"}

COUNT

EXISTS

json

{  "dcl": "1.0",  "action": "EXISTS",  "table": "users",  "where": [    "email = '[email protected]'"  ]}

{  "dcl": "1.0",  "action": "EXISTS",  "table": "users",  "where": [    "email = '[email protected]'"  ]}

SCHEMA

TABLES

TABLE_CREATE — Create a table

Declares a new empty table. Internally creates a single status=8 schema declaration row.

With optional schema definition:

json

{  "dcl": "1.0",  "action": "TABLE_CREATE",  "table": "users",  "schema": {    "name":  "text",    "age":   "integer",    "email": "text"  }}

{  "dcl": "1.0",  "action": "TABLE_CREATE",  "table": "users",  "schema": {    "name":  "text",    "age":   "integer",    "email": "text"  }}

Without schema: Created with schema: null. Column types are auto-registered as text on first INSERT.

status=8 Schema Declaration Row

DataClawe maintains one status=8 declaration row per table.

Example status=8 row content:

Auto Schema Evolution:

Writing to a non-existent table via INSERT auto-creates the status=8 row. New columns are automatically appended to the status=8 schema on each INSERT (type defaults to text).

First INSERT: {"name": "kimura", "age": 25}  → Auto-create status=8: {"name":"text", "age":"text"}  → INSERT the record
Second INSERT: {"name": "suzuki", "phone": "090-xxxx"}  → "phone" is new → auto-append to status=8 schema  → schema: {"name":"text", "age":"text", "phone":"text"}

First INSERT: {"name": "kimura", "age": 25}  → Auto-create status=8: {"name":"text", "age":"text"}  → INSERT the record
Second INSERT: {"name": "suzuki", "phone": "090-xxxx"}  → "phone" is new → auto-append to status=8 schema  → schema: {"name":"text", "age":"text", "phone":"text"}

When connecting to legacy MySQL/PostgreSQL, actual column types (VARCHAR(255), INT, etc.) are recorded in the status=8 row.

STATS — Table statistics

Returns record count, last updated timestamp, and storage size. Useful for AI agents to assess table state before operating.

Example response:

json

{  "dcl": "1.0",  "status": "OK",  "data": {    "table": "users",    "record_count": 1024,    "last_updated": "2026-03-28T10:00:00Z",    "size_kb": 512  }}

{  "dcl": "1.0",  "status": "OK",  "data": {    "table": "users",    "record_count": 1024,    "last_updated": "2026-03-28T10:00:00Z",    "size_kb": 512  }}

TIMELINE — Record change history

Returns the change history (created, updated, logical delete) of a record in chronological order. Useful for AI agents tracking data evolution.

TABLE_COPY — Copy a table

Copies a table under a new name. Used for backups, migrations, and testing.

json

{  "dcl": "1.0",  "action": "TABLE_COPY",  "table": "users",  "target_table": "users_backup_20260328"}

{  "dcl": "1.0",  "action": "TABLE_COPY",  "table": "users",  "target_table": "users_backup_20260328"}

TABLE_RENAME — Rename a table

Renames a table.

TABLE_DROP — Drop a table (logical delete)

Logically deletes a table. Data is not physically removed immediately.

Note: TABLE_DROP is a logical delete — not equivalent to SQL DROP TABLE. Data is not immediately destroyed.

Response Format

Success

json

{  "dcl": "1.0",  "status": "OK",  "count": 42,  "data": [    { "id": 1, "name": "kimura", "email": "[email protected]" }  ],  "meta": {    "table": "users",    "elapsed_ms": 12  }}

{  "dcl": "1.0",  "status": "OK",  "count": 42,  "data": [    { "id": 1, "name": "kimura", "email": "[email protected]" }  ],  "meta": {    "table": "users",    "elapsed_ms": 12  }}

Error

json

{  "dcl": "1.0",  "status": "ERROR",  "code": "TABLE_NOT_FOUND",  "message": "Table 'users' does not exist"}

{  "dcl": "1.0",  "status": "ERROR",  "code": "TABLE_NOT_FOUND",  "message": "Table 'users' does not exist"}

Error codes

Code	Description
`TABLE_NOT_FOUND`	Specified table does not exist
`COLUMN_NOT_FOUND`	Specified column does not exist
`INVALID_ACTION`	Unknown action specified
`INVALID_WHERE`	WHERE condition could not be parsed
`PERMISSION_DENIED`	Tenant does not have access
`CONNECTION_ERROR`	Database connection failed

Integration

DCL works over two protocols. The DCL command itself is identical in both cases.

DCL payload (identical in both protocols)

json

{  "dcl": "1.0",  "action": "SELECT",  "table": "orders",  "where": [    "status = 'pending'",    "created_at >= '2026-01-01'"  ],  "order": "created_at DESC",  "limit": 100}

{  "dcl": "1.0",  "action": "SELECT",  "table": "orders",  "where": [    "status = 'pending'",    "created_at >= '2026-01-01'"  ],  "order": "created_at DESC",  "limit": 100}

This DCL JSON is the same regardless of whether you call via REST API or MCP. Only the outer protocol layer differs.

Supported Databases

DCL normalizes differences between MySQL and PostgreSQL. You write one DCL command. DataClawe handles the rest.

Complexity Layers

DCL is designed in two layers. You choose the layer you need.

DCL Standard — this specification

For AI agents, frontend engineers, and anyone doing straightforward data operations.

No JOIN. No subquery.
Readable by anyone who knows MySQL basics.
Covers the vast majority of real-world CRUD needs.

If Standard covers your needs, you never have to go further.

DCL Advanced — coming in v1.1

For backend engineers who need more expressive power.

Same JSON structure. Same DataClawe engine. More capability when you need it.

JOIN

json

{  "dcl": "1.0",  "action": "SELECT",  "table": "users",  "columns": ["users.id", "users.name", "orders.amount"],  "join": [    {      "table": "orders",      "on": "users.id = orders.user_id",      "type": "LEFT"    }  ],  "where": [    "users.status = 'active'"  ],  "order": "orders.amount DESC",  "limit": 20}

{  "dcl": "1.0",  "action": "SELECT",  "table": "users",  "columns": ["users.id", "users.name", "orders.amount"],  "join": [    {      "table": "orders",      "on": "users.id = orders.user_id",      "type": "LEFT"    }  ],  "where": [    "users.status = 'active'"  ],  "order": "orders.amount DESC",  "limit": 20}

Subquery

json

{  "dcl": "1.0",  "action": "SELECT",  "table": "users",  "where": [    "id IN (SELECT user_id FROM orders WHERE status = 'paid')"  ]}

{  "dcl": "1.0",  "action": "SELECT",  "table": "users",  "where": [    "id IN (SELECT user_id FROM orders WHERE status = 'paid')"  ]}

Multiple JOINs

json

{  "dcl": "1.0",  "action": "SELECT",  "table": "orders",  "columns": [    "orders.id",    "users.name AS customer",    "products.name AS product",    "orders.amount"  ],  "join": [    {      "table": "users",      "on": "orders.user_id = users.id",      "type": "INNER"    },    {      "table": "products",      "on": "orders.product_id = products.id",      "type": "LEFT"    }  ],  "where": [    "orders.status = 'paid'",    "orders.created_at >= '2026-01-01'"  ],  "order": "orders.created_at DESC"}

{  "dcl": "1.0",  "action": "SELECT",  "table": "orders",  "columns": [    "orders.id",    "users.name AS customer",    "products.name AS product",    "orders.amount"  ],  "join": [    {      "table": "users",      "on": "orders.user_id = users.id",      "type": "INNER"    },    {      "table": "products",      "on": "orders.product_id = products.id",      "type": "LEFT"    }  ],  "where": [    "orders.status = 'paid'",    "orders.created_at >= '2026-01-01'"  ],  "order": "orders.created_at DESC"}

JOIN types supported in Advanced

Why two layers?

SQL solved everything in one spec.That is why SQL is still hard after 30 years.
DCL Standard is for everyone.DCL Advanced is for when you need more.You always start simple. You go deeper only when you must.

SQL solved everything in one spec.That is why SQL is still hard after 30 years.
DCL Standard is for everyone.DCL Advanced is for when you need more.You always start simple. You go deeper only when you must.

The same engine handles both. The same JSON structure. No new syntax to learn when you move from Standard to Advanced — just new keys.

What remains intentionally unsupported

The following are out of scope in both Standard and Advanced, to keep DCL safe and predictable for AI agents:

Stored procedures
Schema creation or ALTER TABLE
Raw SQL passthrough
Physical schema destruction (immediate DROP TABLE equivalent)

Note: TABLE_DROP in DCL is a logical delete — distinct from SQL DROP TABLE. It does not immediately destroy data.

DCL is a data operation language, not a schema management language.

Versioning

The "dcl": "1.0" field is required in every request and response.

Future versions will remain backward compatible. A DCL 1.0 request will always work against a DCL 2.0 server.

Contributing

DCL is an open specification. Feedback, proposals, and pull requests are welcome.

Open an issue to propose a new action or operator
Open a pull request to improve documentation or examples
All contributions must follow the design philosophy: readable, MySQL-familiar, BASIC-level simplicity

Roadmap

v1.0 — DCL Standard

Spec finalization
JSON Schema for validation (dcl.schema.json)
Reference implementation in Go (DataClawe Engine)
MCP server reference implementation
MySQL wire protocol support
Multi-tenant isolation specification

v1.1 — DCL Advanced

JOIN specification (INNER / LEFT / RIGHT)
Subquery support
Nested aggregation

SDKs

JavaScript / TypeScript
PHP
Python
Go

Pricing

DataClawe uses a tiered pricing model that scales from individual developers to enterprise.

Initial registration fee: $20 (one-time, all plans)

Usage rates (Business plan — overage)

Record size limit: 100KB per record. This covers IoT sensor data, CRM contacts, medical text records, WordPress posts, and financial transactions. Images and video files are out of scope — store them in CDN/object storage and keep the URL in DataClawe.

For Enterprise pricing, contact us at dataclawe.com/enterprise.

Background

DCL is developed as part of the DataClawe project.

DataClawe is a database translation engine that connects legacy MySQL and PostgreSQL systems to AI agents, LLM pipelines, and cloud-native services — without rewriting the original system.

DCL is the command language that makes this connection possible for everyone, not just database engineers.

Legacy systems should not be destroyed. They should be translated.

License

DCL specification is released under Creative Commons Attribution 4.0 International (CC BY 4.0).

You are free to implement, extend, and build upon this specification. Attribution to DataClawe is appreciated.

DCL v1.1 — 2026 — DataClawe Project

来源：README.md，提交 373a546

工具

该服务器需要登录，因此 SourceWeft 无法在没有账号的情况下读取其工具，也未收录。已连接账号所能看到的内容可能因账号及其权限而异。

版本历史

v1.1.0最新Sep 16, 2026

评分

已满足适用检查项的 75%

该评分衡量此收录条目的完整程度，依据注册中心条目、端点与公开来源计算得出。它不是对服务器本身的评价，也不表示服务器安全或当前可用：可连通性是带日期的检查结果，评分也不包含安装量或用户评价。

近 30 天内可连通必需25 分
已声明认证方式必需10 分
已列出工具15 分
请让服务器在无凭据时响应 tools/list，或发布可在沙箱中无凭据启动的软件包。
描述必需15 分
图标10 分
文档15 分
许可证10 分
请在仓库或注册中心条目中声明许可证。

status value	Purpose
`status=1`	Normal record (active)
`status=8`	Table declaration + schema cache
`status=9`	Logically deleted

Protocol	Used by	Guide
REST API (HTTPS POST)	Frontend, Backend, any HTTP client	See README.api.md
MCP (JSON-RPC 2.0 / SSE)	AI agents (Claude, GPT, Cursor)	See README.mcp.md

Database	Status
MySQL 5.x / 8.x	✅ Supported
PostgreSQL 9–18	✅ Supported
Others	📋 Planned

Type	Description
`INNER`	Records matching in both tables
`LEFT`	All left, matching right
`RIGHT`	All right, matching left

Plan	Monthly	Records	Sessions	SLA
Free	$0	2,000	100/day	No
Personal	$20	up to 20,000	up to 20,000/month	No
Business	$20+	Unlimited	Unlimited	No
Enterprise	Contact us	Unlimited	Unlimited	Yes

Item	Unit price
Record storage	$0.001 / record / month
Session	$0.001 / session