How to Build a SQL Agent with LangChain

To build a SQL Agent in LangChain, use the SQLDatabaseChain class with a language model and a database connection. This agent lets you ask natural language questions that it converts into SQL queries and returns results.

Syntax

The main components to build a SQL agent in LangChain are:

• Language Model: A model like OpenAI's GPT to understand and generate SQL queries.
• Database Connection: Connect to your SQL database using SQLDatabase.
• SQLDatabaseChain: Combines the language model and database to create the agent.

Basic syntax pattern:

from langchain import OpenAI, SQLDatabase, SQLDatabaseChain

db = SQLDatabase.from_uri(your_database_connection)
llm = OpenAI(temperature=0)

agent = SQLDatabaseChain(llm=llm, database=db, verbose=True)

python

from langchain import OpenAI, SQLDatabase, SQLDatabaseChain

# Connect to your database
# Example: SQLite file path
db = SQLDatabase.from_uri("sqlite:///your_database.db")

# Initialize language model
llm = OpenAI(temperature=0)

# Create SQL agent
agent = SQLDatabaseChain(llm=llm, database=db, verbose=True)

Example

This example shows how to create a SQL agent that answers questions about a SQLite database. It connects to the database, sets up the language model, and runs a query in natural language.

python

from langchain import OpenAI, SQLDatabase, SQLDatabaseChain

# Connect to SQLite database file
db = SQLDatabase.from_uri("sqlite:///example.db")

# Initialize OpenAI language model
llm = OpenAI(temperature=0)

# Create the SQL agent
agent = SQLDatabaseChain(llm=llm, database=db, verbose=True)

# Ask a natural language question
question = "How many users signed up last month?"

# Run the agent
answer = agent.run(question)
print(answer)

Output

Number of users who signed up last month: 42

Common Pitfalls

• Incorrect database URI: Make sure the connection string matches your database type and location.
• Verbose mode off: Without verbose=True, debugging SQL queries is harder.
• Language model temperature too high: Use temperature=0 for deterministic SQL generation.
• Unsupported SQL dialect: Some databases may require specific drivers or dialect support.

python

from langchain import OpenAI, SQLDatabase, SQLDatabaseChain

# Wrong: Missing 'sqlite:///' prefix
# db = SQLDatabase.from_uri("example.db")  # This will fail

# Correct:
db = SQLDatabase.from_uri("sqlite:///example.db")

# Wrong: High temperature causes unpredictable SQL
# llm = OpenAI(temperature=0.7)

# Correct:
llm = OpenAI(temperature=0)

agent = SQLDatabaseChain(llm=llm, database=db, verbose=True)

Quick Reference

Summary tips for building a SQL agent with LangChain:

• Use SQLDatabase.from_uri() to connect your database.
• Set temperature=0 in your language model for reliable SQL.
• Enable verbose=True to see generated SQL queries for debugging.
• Use SQLDatabaseChain to combine the model and database.
• Test queries with simple natural language questions first.

Key Takeaways

Use SQLDatabaseChain with a language model and database connection to build a SQL agent.

Set language model temperature to 0 for consistent SQL query generation.

Always verify your database URI format matches your database type.

Enable verbose mode to debug and view generated SQL queries.

Start with simple natural language questions to test your SQL agent.