PostgreSQL Analytics MCP Server

PostgreSQL Analytics for AI Assistants

Go beyond basic SQL. This MCP server gives your AI assistant deep visibility into PostgreSQL databases: smart schema
exploration, query plan analysis with bottleneck annotations, index optimization suggestions, table health monitoring, referential integrity auditing, and natural language to SQL generation.

Tools

• explore_schema — Structured schema tree: tables, columns, types, FKs, indexes, row counts, sizes
• analyze_query — EXPLAIN ANALYZE with human-readable bottleneck annotations (seq scans, disk spills, estimate mismatches)
• suggest_indexes — Analyze slow queries or entire databases, suggest optimal indexes with CREATE INDEX statements
• table_stats — Row counts, dead tuples, bloat %, vacuum/analyze history, cache hit ratios
• find_missing_fks — Detect *_id columns without FK constraints, auto-generate ALTER TABLE suggestions
• generate_query — Natural language to SQL using actual schema context, with optional execution

Installation

{
  "mcpServers": {
    "postgres": {
      "command": "node",
      "args": ["/path/to/mcp-postgres-analytics/dist/index.js"],                                                           
      "env": {
        "DATABASE_URL": "postgres://user:password@localhost:5432/mydb"                                                     
      }                                                                                                                    
    }
  }                                                                                                                        
}                                                         

Use cases

- Understand a new database in one prompt — full schema with row counts and FK relationships                               
- Optimize slow queries with annotated EXPLAIN ANALYZE output
- Find missing foreign keys across an entire schema in seconds                                                             
                                                                                                                           
License
                                                                                                                           
MIT — full source on archimedes.market                                                                                     
 
---