crewAI/crewai_tools/tools/singlestore_search_tool

SingleStoreSearchTool

Description

The SingleStoreSearchTool is designed to facilitate semantic searches and SQL queries within SingleStore database tables. This tool provides a secure interface for executing SELECT and SHOW queries against SingleStore databases, with built-in connection pooling for optimal performance. It supports various connection methods and allows you to work with specific table subsets within your database.

Installation

To install the crewai_tools package with SingleStore support, execute the following command:

pip install 'crewai[tools]'

Or install with the SingleStore extra for the latest dependencies:

uv sync --extra singlestore

Or install the required dependencies manually:

pip install singlestoredb>=1.12.4 SQLAlchemy>=2.0.40

Features

• 🔒 Secure Query Execution: Only SELECT and SHOW queries are allowed for security
• 🚀 Connection Pooling: Built-in connection pooling for optimal performance
• 📊 Table Subset Support: Work with specific tables or all tables in the database
• 🔧 Flexible Configuration: Multiple connection methods supported
• 🛡️ SSL/TLS Support: Comprehensive SSL configuration options
• ⚡ Efficient Resource Management: Automatic connection lifecycle management

Basic Usage

Simple Connection

from crewai_tools import SingleStoreSearchTool

# Basic connection using host/user/password
tool = SingleStoreSearchTool(
    host='localhost',
    user='your_username',
    password='your_password',
    database='your_database',
    port=3306
)

# Execute a search query
result = tool._run("SELECT * FROM employees WHERE department = 'Engineering' LIMIT 10")
print(result)

Working with Specific Tables

# Initialize tool for specific tables only
tool = SingleStoreSearchTool(
    tables=['employees', 'departments'],  # Only work with these tables
    host='your_host',
    user='your_username',
    password='your_password',
    database='your_database'
)

Complete CrewAI Integration Example

Here's a complete example showing how to use the SingleStoreSearchTool with CrewAI agents and tasks:

from crewai import Agent, Task, Crew
from crewai_tools import SingleStoreSearchTool

# Initialize the SingleStore search tool
singlestore_tool = SingleStoreSearchTool(
    tables=["products", "sales", "customers"],  # Specify the tables you want to search
    host="localhost",
    port=3306,
    user="root",
    password="pass",
    database="crewai",
)

# Create an agent that uses this tool
data_analyst = Agent(
    role="Business Analyst",
    goal="Analyze and answer business questions using SQL data",
    backstory="Expert in interpreting business needs and transforming them into data queries.",
    tools=[singlestore_tool],
    verbose=True,
    embedder={
        "provider": "ollama",
        "config": {
            "model": "nomic-embed-text",
        },
    },
)

# Define a task
task = Task(
    description="List the top 2 customers by total sales amount.",
    agent=data_analyst,
    expected_output="A ranked list of top 2 customers that have the highest total sales amount, including their names and total sales figures.",
)

# Run the crew
crew = Crew(tasks=[task], verbose=True)
result = crew.kickoff()

Advanced CrewAI Example with Multiple Agents

from crewai import Agent, Task, Crew
from crewai_tools import SingleStoreSearchTool

# Initialize the tool with connection URL
singlestore_tool = SingleStoreSearchTool(
    host="user:password@localhost:3306/ecommerce_db",
    tables=["orders", "products", "customers", "order_items"]
)

# Data Analyst Agent
data_analyst = Agent(
    role="Senior Data Analyst",
    goal="Extract insights from database queries and provide data-driven recommendations",
    backstory="You are an experienced data analyst with expertise in SQL and business intelligence.",
    tools=[singlestore_tool],
    verbose=True
)

# Business Intelligence Agent
bi_specialist = Agent(
    role="Business Intelligence Specialist",
    goal="Transform data insights into actionable business recommendations",
    backstory="You specialize in translating complex data analysis into clear business strategies.",
    verbose=True
)

# Define multiple tasks
data_extraction_task = Task(
    description="""
    Analyze the sales data to find:
    1. Top 5 best-selling products by quantity
    2. Monthly sales trends for the last 6 months
    3. Customer segments by purchase frequency
    """,
    agent=data_analyst,
    expected_output="Detailed SQL query results with sales analysis including product rankings, trends, and customer segments."
)

insights_task = Task(
    description="""
    Based on the sales data analysis, provide business recommendations for:
    1. Inventory management for top products
    2. Marketing strategies for different customer segments
    3. Sales forecasting insights
    """,
    agent=bi_specialist,
    expected_output="Strategic business recommendations with actionable insights based on the data analysis.",
    context=[data_extraction_task]
)

# Create and run the crew
analytics_crew = Crew(
    agents=[data_analyst, bi_specialist],
    tasks=[data_extraction_task, insights_task],
    verbose=True
)

result = analytics_crew.kickoff()

Connection Methods

SingleStore supports multiple connection methods. Choose the one that best fits your environment:

1. Standard Connection

tool = SingleStoreSearchTool(
    host='your_host',
    user='your_username',
    password='your_password',
    database='your_database',
    port=3306
)

2. Connection URL (Recommended)

You can use a complete connection URL in the host parameter for simplified configuration:

# Using connection URL in host parameter
tool = SingleStoreSearchTool(
    host='user:password@localhost:3306/database_name'
)

# Or for SingleStore Cloud
tool = SingleStoreSearchTool(
    host='user:password@your_cloud_host:3333/database_name?ssl_disabled=false'
)

3. Environment Variable Configuration

Set the SINGLESTOREDB_URL environment variable and initialize the tool without any connection arguments:

# Set the environment variable
export SINGLESTOREDB_URL="singlestoredb://user:password@localhost:3306/database_name"

# Or for cloud connections
export SINGLESTOREDB_URL="singlestoredb://user:password@your_cloud_host:3333/database_name?ssl_disabled=false"

# No connection arguments needed when using environment variable
tool = SingleStoreSearchTool()

# Or specify only table subset
tool = SingleStoreSearchTool(tables=['employees', 'departments'])

4. Connection with SSL

tool = SingleStoreSearchTool(
    host='your_host',
    user='your_username',
    password='your_password',
    database='your_database',
    ssl_ca='/path/to/ca-cert.pem',
    ssl_cert='/path/to/client-cert.pem',
    ssl_key='/path/to/client-key.pem'
)

5. Advanced Configuration

tool = SingleStoreSearchTool(
    host='your_host',
    user='your_username',
    password='your_password',
    database='your_database',
    # Connection pool settings
    pool_size=10,
    max_overflow=20,
    timeout=60,
    # Advanced options
    charset='utf8mb4',
    autocommit=True,
    connect_timeout=30,
    results_format='tuple',
    # Custom connection attributes
    conn_attrs={
        'program_name': 'MyApp',
        'custom_attr': 'value'
    }
)

Configuration Parameters

Basic Connection Parameters

• host: Database host address or complete connection URL
• user: Database username
• password: Database password
• port: Database port (default: 3306)
• database: Database name
• tables: List of specific tables to work with (optional)

Connection Pool Parameters

• pool_size: Maximum number of connections in the pool (default: 5)
• max_overflow: Maximum overflow connections beyond pool_size (default: 10)
• timeout: Connection timeout in seconds (default: 30)

SSL/TLS Parameters

• ssl_key: Path to client private key file
• ssl_cert: Path to client certificate file
• ssl_ca: Path to certificate authority file
• ssl_disabled: Disable SSL (default: None)
• ssl_verify_cert: Verify server certificate
• ssl_verify_identity: Verify server identity

Advanced Parameters

• charset: Character set for the connection
• autocommit: Enable autocommit mode
• connect_timeout: Connection timeout in seconds
• results_format: Format for query results ('tuple', 'dict', etc.)
• vector_data_format: Format for vector data ('binary', 'json')
• parse_json: Parse JSON columns automatically

For more detailed connection options and advanced configurations, refer to the SingleStore Python SDK documentation.