Here’s why it qualifies as an SQL Agent and how it works:
1️.Accepts natural language input (e.g., “Show all employees in the sales department”).
2️.Converts it into a valid SQL query using Groq’s LLM (via LangChain).
3️.Dynamically fetches table schema (columns & types) from MySQL for accurate SQL generation.
4️.Injects this schema into the LLM prompt for context-aware queries.
5️.Differentiates between raw SQL and NL input, handling both seamlessly.
6️.Executes SELECT, INSERT, UPDATE, DELETE on the database.
7️.Cleans up LLM output by removing markdown/code artifacts like ```sql.
8️.Exposes everything through a FastAPI endpoint, making it easy for clients to query via API.
#How to run locally cd sql-agent pip install -r requirements.txt
pip show uvicorn pip install uvicorn[standard] pip install fastapi uvicorn psycopg2-binary pip install openai langchain langchain-community
to start interactive shell python sql_shell.py Run the client script in another terminal:python client.py pip install langchain langchain-openai pip install python-dotenv https://console.groq.com/home[APIKEY]
uvicorn app:app --reload --port 8000
curl -X POST http://localhost:8000/query
-H "Content-Type: application/json"
-d '{"sql":"SELECT NOW() as ts;"}'
Build & “publish” with Docker
docker build -t sql-agent:latest .
docker run --rm -p 8000:8000
-e PG_HOST=host.docker.internal
-e PG_PORT=5432
-e PG_DB=postgres
-e PG_USER=postgres
-e PG_PASS=postgres
sql-agent:latest
curl -X POST http://localhost:8000/sqlagent
curl -X 'POST'
'http://127.0.0.1:8000/sqlagent'
-H 'accept: application/json'
-H 'Content-Type: application/json'
-d '{
"sql": "change email to jayaai7650@gmail.com all fields fields of employee "
}
How to get token: https://console.groq.com/home just signup and get details.
curl -X 'POST'
'http://127.0.0.1:8000/sqlagent'
-H 'accept: application/json'
-H 'Content-Type: application/json'
-d '{
"sql": "add data to managers table id 100216, manager_name jaya"
}'
Response body Download { "query": "INSERT INTO employee (EmpID, FirstName, LastName, Email, PhoneNumber, HireDate, JobTitle, Salary, DepartmentID) \nVALUES (100216, 'Jaya', NULL, NULL, NULL, NULL, 'Manager', NULL, NULL);", "rows": [ { "status": "success", "affected_rows": 1 } ] }
About Me I’m Jayakrishna M, an IT professional with 10+ years of experience. I began my career in Java & Spring Boot and have since expanded into Python, LLMs, AI Agents, NumPy, and Pandas. Recently, I built my first SQL Agent project.
📩 Feel free to reach out: mjk7650@gmail.com