Works with the AI Agents you already use

    Claude CodeClaude CodeCursorCursorCodex CLICodex CLIMuseMuseDotsDots+21 more

    MySQL/PostgreSQL MCP Server

    1

    Connect agents to MySQL/MariaDB or PostgreSQL via MCP for safe schema discovery, paginated queries, catalog inspection, and guarded writes.

    Free

    1 useSecurity scanned
    MySQL/PostgreSQL MCP Server

    MySQL/PostgreSQL MCP Server

    Example session with this skill installed

    Show me the tables in the 'inventory' database and describe the 'products' table.

    • Read your context and instructions
    • Compiled the mysql postgresql mcp

    I have retrieved the schema information for the requested database.

    MySQL summary

    • Database: inventory
    • Mode: read-only
    • What ran: mysql_query to show tables and describe products

    Findings

    The inventory database contains four tables. The products table stores core item data including pricing and stock levels, using an auto-incrementing integer as the primary key.

    Data sample

    Table NameType
    categoriesBASE TABLE
    productsBASE TABLE
    stock_levelsBASE TABLE
    suppliersBASE TABLE

    Structure of products:

    FieldTypeNullKeyDefault
    idint(11)NOPRINULL
    skuvarchar(50)NOUNINULL
    namevarchar(255)NONULL
    pricedecimal(10,2)NO0.00
    statusenum('active','deleted')YES'active'

    Next steps

    • Run a SELECT query on products with a limit to see sample data.
    • Use mysql_catalog if you need to see triggers associated with the products table.
    • Verify if MYSQL_PII_COLUMNS is configured if you plan to query customer-linked data.

    Connects securely to your tools. The creator never sees your data.

    About this skill

    Interacting with MySQL, MariaDB, or PostgreSQL from an LLM chat often means risky shell commands or exposed credentials. This skill gives agents a structured workflow for @eduartedev/mysql-mcp-server: configure MCP once, discover schema first, then run paginated queries with strict security defaults.

    What it does

    • Multi-engine MCP — one binary for MySQL/MariaDB (mysql_query / mysql_catalog) or PostgreSQL (postgres_query / postgres_catalog), selected with DB_ENGINE or port autodetect.
    • Schema discovery — map tables/columns (SHOW/DESCRIBE or Postgres equivalents) before writing complex SQL.
    • Safe querying — read-only by default, with result caps to protect the context window.
    • Automated pagination — follow hasMore / nextOffset across pages (limit default 20, max 100).
    • Logic inspection — read views, triggers, functions, and procedures via catalog tools.
    • Write guardrails — INSERT/UPDATE/DELETE only when DB_ACCESS_MODE=readwrite, with mandatory real WHERE checks and allowlists (:ro / :rw).
    • PII masking & TLS — sensitive columns returned as ***; optional DB_SSL=required for RDS and other TLS setups.

    How it works

    1. Configure MCP — add the stdio server (npx -y @eduartedev/mysql-mcp-server) with DB_* env (host, credentials, allowlist, optional SSL).
    2. Discover scope — list tables/schemas and describe columns before querying.
    3. Execute and paginate — run validated SQL and continue with nextOffset while hasMore is true.

    Frameworks & tools

    Built for @eduartedev/mysql-mcp-server on Node.js 18+. Works with MCP stdio clients such as Cursor, Claude Code, and Codex CLI.

    Why this beats prompting it yourself

    Generic prompts push agents toward mysql/psql shells or invented schemas. This skill enforces discovery-then-query, PII masking, blocked DDL/admin commands, and controlled writes.

    Use cases

    • Debug row-level issues with a least-privilege, readonly DB user.
    • Audit database logic by reading trigger/function/procedure definitions.
    • Run safe, paginated exports for local analysis or reporting.
    • Query MySQL locally or PostgreSQL on RDS with TLS + CA bundle.

    Known limitations

    Requires Node.js 18+ and an MCP client with stdio. No multi-statement queries, no DDL (DROP/ALTER/CREATE/…), no LOAD DATA / admin paths like GRANT. Writes stay off until you set DB_ACCESS_MODE=readwrite and restart MCP.

    Requires Node.js 18+ and an MCP client with stdio support. No support for multi-statement queries, DDL operations, or administrative commands like GRANT.

    How to use

    1. 1

      Connect your agent to the Agensi MCP

      Copy the prompt and paste it into Claude, ChatGPT, Cursor or any agent that supports MCP. It sets up the connection for you. Manual setup

    2. 2

      Sign in to Agensi

      Your agent shows a sign-in link. Sign in and click Allow.

    3. 3

      Ask for what you need

      Your agent loads this skill and uses it on your task. It always asks before any payment.

    Reviews

    No reviews yet

    Be one of the first to try it. Every listed skill goes through the trust checks below.

    Security scanned

    Checked by our 8-point scan before listing

    1 use

    Times used to date

    Free forever

    No account required to browse

    Trust & safety

    Security scan

    Security scan passed 5 days ago

    • Free with an Agensi account

    Requires

    @eduartedev/mysql-mcp-server
    Node.js
    Enabled MCP
    Config MCP

    Permissions

    Network Access
    Listed27 days ago
    Updated5 days ago

    What's inside

    Frequently Asked Questions

    This skill bridges the gap between LLMs and your MySQL/MariaDB databases, allowing agents to perform schema discovery and paginated queries safely through the Model Context Protocol (MCP) rather than risky shell commands.

    The skill is compatible with any MCP-compliant client that supports stdio, including popular development tools like Cursor, Claude Code, and Codex CLI.

    Installation requires Node.js 18+; you simply add the server to your MCP client settings using the provided stdio configuration and your specific database environment variables.

    By default the MCP runs in read-only mode. Writes (INSERT, UPDATE, DELETE) only work when you set MYSQL_ACCESS_MODE=readwrite and restart MCP. UPDATE and DELETE must include a meaningful WHERE clause; always-true conditions are rejected

    No, for security and stability, this skill specifically blocks DDL commands such as DROP or ALTER, as well as multi-statement queries and file system operations like LOAD DATA.

    New in Data & Databases