Quick Start¶
This page used to repeat Getting Started. Its sections now point to the one place that keeps each step, and it keeps the recorded walkthroughs.
- New to CUBRID? Start with the beginner tutorial,
your first CUBRID database from Python: five short lessons
from a local practice database to a first query,
commit,rollbackand a safe clean-up. - Already know which path you want? Use Getting Started.
0. Start CUBRID (all paths)¶
Starting the lab and waiting until it is ready: lesson 1 of the tutorial, Prepare the lab, or Getting Started.
1. Five-minute first query (pycubrid)¶
See Getting Started. A guided version is Lesson 2: Run your first query.
2. ORM and application templates (sqlalchemy-cubrid)¶
See Getting Started and the SQLAlchemy quickstart.
3. Natural language over the database (cubrid-mcp-server)¶
See Getting Started.
Recorded walkthroughs¶
Screen recordings of the examples running against CUBRID. Every command and
every line of output in them is real; the waits for the CUBRID health check and
the dashboard's first pip install are cut. How they were made, with versions
and hashes: demos/.
Streamlit dashboard: one command¶
docker compose up -d in templates/dashboard, then the table viewer at
http://localhost:8501 (CUBRID 11.4).
Download the dashboard walkthrough.
AI agent state management¶
templates/ai-agent/01_agent_state.py stores a session, its messages and a tool
call in CUBRID; 04_agent_loop.py runs a query, think, act, observe loop and
persists every turn. The planner is a heuristic, not an LLM.
Download the agent state walkthrough.
MCP tool chain¶
templates/ai-agent/02_mcp_toolchain.py starts cubrid-mcp-server over stdio,
lists its tools, calls them and shows the read-only whitelist rejecting a
DROP TABLE. No LLM client or API key is involved.
Download the MCP tool chain walkthrough.
Verify a full checkout¶
See Getting Started.