mcp-win-stdio-db

Unified Database Model Context Protocol (MCP) server for PostgreSQL and MySQL with multi-server in-memory pooling, sibling database auto-derivation, cross-schema resolution, and complete DBA operations.

Part of the mcp-win-stdio Windows-optimized suite.

🚀 Features (31 Tools)

• Polyglot Database Engine: Handles PostgreSQL (psycopg2) and MySQL (pymysql) transparently side-by-side.
• Entity-Relationship Diagrams (generate_erd): Generates clean GitHub/Mermaid markdown ER diagrams from foreign key metadata.
• Row-Level Data Diffing (diff_data): Compares data across tables, schemas, or connections with column-by-column diffs.
• Performance & Lock Auditing (list_slow_queries, get_locks): Real-time blocking dependency trees and slow query analysis (pg_stat_statements / performance_schema).
• Streaming Export & Batch Import (export_table, import_csv): Streams tables/queries to CSV, JSON, or TSV on disk, and safely batch imports CSVs with transaction dry-run simulation.
• Database Health & Schema Diff: Unindexed foreign keys, unused indexes, bloated tables with recommended SQL fixes (audit_database_health, compare_schemas).
• In-Memory Connection Pooling: Fast continuous queries with zero reconnection latency and persistent use_database sticky state.
• Detailed Diagnostic Error Reporting: Native error codes (e.g. 23503 FK violation, 23505 unique violation), offending constraint names, and suggestions.
• DBA & Management Tools: create_database, drop_database (with safety guard), clone_database (instant template clone), terminate_connections, dump_database, and restore_database.

🛠️ Included Tools (31 Tools)

1. list_connections: Lists all active and configured database connections.
2. use_database: Switches active connection or switches database on current server.
3. list_databases: Lists all databases on the active server.
4. list_schemas: Lists all schemas in the active PostgreSQL database.
5. describe_table: Detailed schema inspection (columns, nullability, PKs, FKs, indexes). Auto-resolves cross-schema tables in PostgreSQL.
6. schema_overview: Compact overview of all user tables and views.
7. compact_schema_overview: Ultra-compact one-liner per table: schema.table (col: type PK, col2: FK->ref).
8. get_table_sample: Returns sample rows and estimated row count with PII masking.
9. search_schema: Case-insensitive regex search for table, view, or column names.
10. get_table_ddl: Complete CREATE TABLE DDL reconstruction.
11. get_database_stats: Table sizes, row estimates, index sizes, and total database size.
12. read_query: Safe SELECT queries returning structured JSON with row counts.
13. execute_query: Executes DML / DDL statements with transaction commit/rollback and dry_run support.
14. explain_query: Generates EXPLAIN / EXPLAIN ANALYZE execution plan.
15. list_active_queries: Lists running queries, connection duration, and process IDs.
16. analyze_table_indexes: Table index analysis: PKs, missing FK indexes, unused indexes.
17. compare_schemas: Schema diff between databases/schemas with generated migration SQL.
18. audit_database_health: Unindexed FKs, unused indexes, bloated tables with SQL fixes.
19. generate_erd: Entity-Relationship Diagram in GitHub/Mermaid markdown syntax from foreign keys.
20. diff_data: Row-level data comparison with field-level diffs and key matching.
21. list_slow_queries: Slow query analysis (pg_stat_statements / performance_schema).
22. get_locks: Real-time blocking dependency trees and lock distributions.
23. export_table: Streams table contents or SQL query results directly to CSV, JSON, or TSV on disk.
24. import_csv: Batch imports CSV file into table with transaction safety and dry-run rollback.
25. add_connection: Dynamically adds a new PostgreSQL or MySQL connection string at runtime.
26. create_database: Creates a new database on the active server.
27. drop_database: Safety-guarded DROP DATABASE (requires confirmName matching target name).
28. clone_database: Fast database cloning using PostgreSQL TEMPLATE mechanism.
29. terminate_connections: Terminates active connections to a specific database.
30. dump_database: Dumps database to SQL file using native pg_dump or mysqldump.
31. restore_database: Restores database from SQL dump file using native psql or mysql.

📦 Installation

pip install mcp-win-stdio-db

(Installing this package automatically installs mws CLI orchestrator).

🚀 One-Command Claude Setup

mws setup db
# or:
mws add db

Manual Configuration Example

In %APPDATA%\Claude\claude_desktop_config.json:

{
  "mcpServers": {
    "db": {
      "command": "python",
      "args": ["-m", "mcp_win_stdio.db"],
      "env": {
        "SERVERS": "{\"postgres\":\"postgresql://postgres:password@localhost:5432/mydb\",\"mysql\":\"mysql://user:pass@localhost:3306/mydb\"}"
      }
    }
  }
}

📖 CLI Commands & Interactive Guide

mws db guide       # Complete tool reference & prompt recipes
mws db doctor      # Verify PostgreSQL/MySQL adapters and native dump tools
mws db setup       # Configure Claude Desktop / Claude Code
mws db run         # Launch server over stdio

📜 License

MIT License. Copyright (c) 2026 Mohan Kumar Indala.