Model Context Protocol moderate risk

Alchemy

MCP Alchemy connects Claude to SQL databases, enabling exploration, query execution, and data analysis, with configurable access and connection pooling.

Connections & Capabilities

Connects To

GitHubPostgreSQLMySQLSQLiteAzure

Capabilities

readwriteexec

Quickstart

Config

{
  "mcpServers": {
    "my_sqlite_db": {
      "command": "uvx",
      "args": ["--from", "mcp-alchemy==2025.8.15.91819",
               "--refresh-package", "mcp-alchemy", "mcp-alchemy"],
      "env": {
        "DB_URL": "sqlite:////absolute/path/to/database.db"
      }
    }
  }
}

Exposed MCP Tools (4)

safe

all_table_names

Retrieves all table names in the connected database.

Read-only operation; no data modification.

safe

filter_table_names

Finds table names matching a specified substring.

Read-only operation; no data modification.

safe

schema_definitions

Gets detailed schema information for specified tables.

Read-only operation; no data modification.

high

execute_query

Executes an arbitrary SQL query against the database.

Can modify, delete, or expose sensitive data based on the query.

Safety Assessment

MCP Alchemy's safety depends on the user's SQL query practices. Read-only operations are relatively safe. However, executing arbitrary SQL queries can be risky if not properly sanitized, potentially leading to data breaches or modifications. Limit access and monitor query execution.

Connection pooling helps manage database resources.
Supports read-only operations for safer data exploration.
Provides options to configure database engine parameters.
Allows setting maximum output length to prevent large data leaks.
Uses pre-ping to validate connections before use.

Executes arbitrary SQL queries, posing a risk of data modification or deletion.
Requires database credentials to be stored as environment variables.
No built-in input sanitization for SQL queries.
Potential for SQL injection if queries are not carefully constructed.
Direct database access without sandboxing.