cubrid-mcp-server

A Model Context Protocol server for the CUBRID database. It lets LLM clients such as Claude Desktop, Claude Code, and Cursor safely inspect schemas and run read-only queries through the pure-Python pycubrid driver.

uvx cubrid-mcp-server

Why this server

• Read-only by default. A code-level SQL whitelist allows only SELECT, SHOW, DESC, DESCRIBE, and WITH statements; multi-statement input is rejected; execution plans come from the explain_query tool. Write access is a separate, opt-in tool.
• Schema-aware. Twelve tools cover table discovery, column metadata, indexes, serials, class hierarchy, row counts, execution plans, and health checks — plus the same metadata as MCP Resources.
• Multi-database. One server process can serve several CUBRID databases, each with independent read-only, write, and audit settings.
• Operations-friendly. All logging goes to stderr (stdout is reserved for the MCP protocol stream), errors returned to the client are sanitized, and an opt-in audit log records every executed statement without ever logging raw SQL.

Tool surface

Tool	Description
all_table_names	List every user table in the database
filter_table_names	Substring search over table names
schema_definitions	Column types, nullability, defaults, and primary key info
describe_table	Full metadata: columns, primary key, and indexes in one call
list_indexes	Indexes for a table with key columns and flags
explain_query	Execution plan/trace for a SELECT/WITH (via CUBRID SHOW TRACE)
table_row_counts	COUNT(*) for one or many tables
list_serials	CUBRID SERIAL sequences with current value and bounds
list_class_hierarchy	CUBRID CLASS inheritance relationships
execute_query	Run read-only SQL with automatic output truncation
health_check	Verify database connectivity on demand
execute_write	Single INSERT/UPDATE/DELETE in an atomic transaction (only registered when opt-in write mode is enabled)

See the full Tools reference for parameters and return shapes, the Security Model for the safety design, and Multi-Connection for serving several databases from one process.

Documentation

• Start here: Quick Start — a disposable lab, a read-only account, one MCP client and a first query, then a safe clean-up
• How-to guides: Serve several databases — named connections and per-connection settings; chat-client setup is at the end of the Quick Start
• Reference: Tools, resources and prompts, configuration settings and the Security Model — read-only enforcement, opt-in writes, audit logging
• Help: Troubleshooting; questions and bugs through the issue forms (never paste passwords, connection strings or query results you would not publish); vulnerabilities privately as described in SECURITY.md
• 한국어: 한국어 README, 빠른 시작, 도구 참조, 보안 모델, 멀티커넥션, 문제 해결

License

MIT — see LICENSE.

This project is part of CUBRID Lab, an independent open-source initiative for CUBRID developer tooling, and is not affiliated with, sponsored by, or endorsed by CUBRID Corporation or the official CUBRID project.

Ecosystem

Part of the cubrid-lab Python ecosystem:

• cubrid-mcp-server — MCP server — natural-language access for LLM clients
• pycubrid — Pure-Python DB-API 2.0 driver for CUBRID (sync + native asyncio)
• sqlalchemy-cubrid — SQLAlchemy 2.0–2.2 dialect + Alembic
• cubrid-cookbook-python — 68 runnable examples and application templates