Text2sql MCP

Synechron Text2SQL MCP服务器

Synechron的Text2SQL MCP服务器是 模型上下文协议 （MCP） 服务器。它为Cursor、VSCode和Claude Desktop等MCP客户端提供了对关系数据库的自然语言访问 各种类型的数据库，使用户能够使用自然语言功能对其数据提出问题。

服务器通过结合使用模式和行感知语义来优化其生成的结果的质量 搜索。模式和行样本被索引，使服务器能够更有效地生成合适的查询。

特性和功能

Text2SQL MCP服务器提供以下功能：

• SQL的自然语言： 使用高级SQL查询将自然语言问题转换为优化的SQL查询

语言模型。

• 多数据库支持： 与PostgreSQL、MySQL、SQLite和Microsoft Fabric数据库兼容。
• 检索增强生成（RAG）： 通过结合模式元数据和采样行增强查询生成

内容。

• 表索引： 支持嵌入数据库模式和示例数据，提高查询相关性。
• 响应综合： 生成查询结果的自然语言摘要。
• Markdown格式： 创建更易读的响应。
• 多型号支持： 适用于OpenAI、Azure OpenAI、AWS Bedrock和Ollama模型。

样品数据

服务器附带了两个示例数据集。

• 网络威胁
• ESG评级

示例客户端

还包括一个示例REPL。

______________________________________________________________________

Docker快速入门

注意：要使用自己的数据库，需要进一步配置，请参阅 配置 章节 在......下面

使用Docker运行服务器

docker run -it -p8000:8000 \
  -v ~/.aws:/home/appuser/.aws:ro \
  -e MODEL_API_TYPE=Bedrock \
  -e AWS_PROFILE=AWSAdministratorAccess-880502554482 \
  -e FASTMCP_HOST=0.0.0.0 \
  709825985650.dkr.ecr.us-east-1.amazonaws.com/synechron/mcp-text2sql:0.0.10-2025.07.04-rc3 server

使用Docker运行示例客户端

docker run -it \
    -v ~/.aws:/home/appuser/.aws:ro \
    -e AWS_PROFILE=AWSAdministratorAccess-880502554482 \
    -e BEDROCK_MODEL_ID=us.anthropic.claude-3-7-sonnet-20250219-v1:0 \
    -e BEDROCK_API_VERSION=2025-01-01-preview \
    -e MCP_SERVER_HOST=host.docker.internal \
     709825985650.dkr.ecr.us-east-1.amazonaws.com/synechron/mcp-text2sql:0.0.10-2025.07.04-rc3 client

客户端使用示例

Connected to server with tools: ['Text2Sql_Cyber_Threats', 'Text2Sql_ESG_Ratings']

MCP Client Started!
Type your queries or 'quit' to exit.

Query: What kind of data is in the Cyber Threat datasource?

The Cyber Threat datasource contains comprehensive information about cybersecurity incidents with the following data fields:

1. country - The country where the cyber attack took place
2. year - The year when the cyber attack occurred
3. attack_type - The type of cyber attack (e.g., Phishing, Ransomware, DDoS, Man-in-the-Middle, SQL Injection)
4. target_industry - The industry that was targeted (e.g., Education, Retail, IT, Telecommunications, Healthcare, Government, Banking)
5. financial_loss_(in_million_$) - The financial impact of the attack in millions of dollars
6. number_of_affected_users - How many users were affected by the attack
7. attack_source - Where the attack originated from (e.g., Hacker Group, Nation-state, Insider, Unknown)
8. security_vulnerability_type - The type of security vulnerability exploited (e.g., Unpatched Software, Weak Passwords, Social Engineering)
9. defense_mechanism_used - What defense was in place (e.g., VPN, Firewall, Antivirus, AI-based Detection)
10. incident_resolution_time_(in_hours) - How long it took to resolve the incident in hours

The database tracks cybersecurity incidents across different countries, industries, and years, including details about attack methods, financial impact, affected users, vulnerabilities exploited, and resolution times.

将服务器与现有MCP客户端一起使用

服务器应该使用Docker运行。使用以下任一配置将其添加到MCP客户端：

{
  "servers": {
    "text2sql": {
      "type": "streamable-http",
      "url": "http://localhost:8000/mcp"
    }
  }
}

{
  "mcpServers": {
    "text2sql": {
      "command": "docker",
      "args": [
        "run",
        "-i",
        "--rm",
        "--env-file",
        ".env",
        "709825985650.dkr.ecr.us-east-1.amazonaws.com/synechron/mcp-text2sql:0.0.10-2025.07.04-rc3",
        "server"
      ]
    }
  }
}

配置

Text2SQL MCP服务器可以配置环境变量和JSON配置的组合。

服务器配置

示例.env文件

LOG_LEVEL=INFO
PYTHONUNBUFFERED=1
TEXT2SQL_VALVES_JSON=/home/appuser/config/config.json

MODE=shttp
FASTMCP_HOST=0.0.0.0
FASTMCP_PORT=8000

AUTH_ENABLED=true
AUTH_JWKS_URI=https://cognito-idp.us-east-1.amazonaws.com/us-east-1_/.well-known/jwks.json
AUTH_ISSUER=https://cognito-idp.us-east-1.amazonaws.com/us-east-1_

AWS_PROFILE=default

示例TEXT2SQL_VALVES_JSON（阀门.JSON）

Text2SQL MCP服务器必须使用指定数据库连接、模型设置和 其他 参数。以下是一个配置示例：

{
  "ENABLED": true,
  "ENABLE_RAG": false,
  "ENABLE_MARKDOWN": true,
  "ENABLE_PANDAS": true,
  "ENABLE_TABLE_INDEXING": true,
  "FORCE_REINDEXING": false,
  "DATABASE_CONFIG": {
    "Cyber_Threats": {
      "description": "A comprehensive dataset tracking cybersecurity incidents, attack vectors, threat types, and affected countries.",
      "topics": [
        "Cyber Threats",
        "Attacks",
        "Targets"
      ],
      "url": "sqlite:///file:/home/appuser/config/cyber_threats.db?uri=true",
      "tables": {
        "cyber_threats": "The Global Cybersecurity Threats Dataset (2015-2024) provides extensive data on cyberattacks, malware types, targeted industries, and affected countries."
      },
      "prompts": [
        {
          "name": "biggest_cyber_threat",
          "description": "Which type of cyber threat has caused the biggest financial loss?"
        }
      ]
    },
    "ESG_Ratings": {
      "description": "S&P 500 Companies ESG Insights & Risk Scores for Informed Decisions.",
      "topics": [
        "ESG Ratings",
        "ESG Risk",
        "Sustainability"
      ],
      "url": "sqlite:///file:/home/appuser/config/esg_ratings.db?uri=true",
      "tables": {
        "esg_ratings": "A comprehensive dataset tracking ESG ratings for S&P 500 companies."
      }
    }
  },
  "IGNORE_SCHEMA": "information_schema, INFORMATION_SCHEMA, _rsc, db_accessadmin, db_backupoperator, db_datareader, db_datawriter, db_ddladmin, db_denydatareader, db_denydatawriter, db_owner, db_securityadmin, guest, queryinsights, sys, pg_catalog",
  "MODEL_API_TYPE": "Bedrock",
  "AWS_PROFILE": "default",
  "API_VERSION": "2025-01-01-preview",
  "LLM_MODEL_NAME": "us.anthropic.claude-3-7-sonnet-20250219-v1:0",
  "EMBED_MODEL_NAME": "amazon.titan-embed-text-v2:0",
  "MAX_RETRY": 3,
  "MAX_RESULTS": 100,
  "MAX_DATAFRAME": 2000,
  "LOG_LEVEL": "INFO"
}

客户端配置

以下环境变量可以提供给示例MCP客户端：

工具

Text2SQL MCP服务器为配置的数据库动态创建MCP工具和提示，格式如下：

• Text2Sql_数据库名称：使用自然语言查询特定数据库

- 参数： - query：对数据库执行的自然语言查询

许可证

该项目根据MIT许可证获得许可。