analyzing-database-indexes

name	analyzing-database-indexes
description	Process use when you need to work with database indexing. This skill provides index design and optimization with comprehensive guidance and automation. Trigger with phrases like "create indexes", "optimize indexes", or "improve query performance".
allowed-tools	Read, Write, Edit, Grep, Glob, Bash(psql:*), Bash(mysql:*), Bash(mongosh:*)
version	1.27.0
author	Jeremy Longshore <[e-mail]>
license	MIT
tags	["database","performance","analyzing-database"]
compatibility	Designed for Claude Code

Database Index Advisor

Overview

Analyze database index usage, identify missing indexes causing sequential scans, detect redundant or unused indexes wasting write performance, and recommend optimal index configurations for PostgreSQL and MySQL.

Prerequisites

• Database credentials with access to pg_stat_user_indexes, pg_stat_user_tables, and pg_stat_statements (PostgreSQL) or performance_schema and sys schema (MySQL)
• pg_stat_statements extension enabled for PostgreSQL query statistics
• psql or mysql CLI for executing analysis queries
• Representative workload running (analysis during off-peak hours may miss important query patterns)
• At least 24 hours of statistics accumulation since the last pg_stat_reset()

Instructions

1. Identify tables with high sequential scan activity (candidates for missing indexes):

   • PostgreSQL: SELECT relname, seq_scan, seq_tup_read, idx_scan, n_live_tup FROM pg_stat_user_tables WHERE seq_scan > 100 AND n_live_tup > 10000 ORDER BY seq_tup_read DESC LIMIT 20
   • A table with high seq_scan count and high seq_tup_read relative to n_live_tup is scanning most of the table repeatedly

2. Find the queries causing sequential scans by correlating with pg_stat_statements:

   • SELECT query, calls, mean_exec_time, rows FROM pg_stat_statements WHERE query ILIKE '%table_name%' ORDER BY mean_exec_time DESC LIMIT 10
   • Run EXPLAIN (ANALYZE, BUFFERS) on the top queries to confirm sequential scan usage

3. Analyze query WHERE clauses and JOIN conditions to determine which columns need indexes. Extract the filtering columns and their selectivity:

   • SELECT column_name, n_distinct, correlation FROM pg_stats WHERE tablename = 'target_table'
   • High n_distinct (close to row count) indicates good index selectivity
   • correlation close to 1.0 or -1.0 suggests the column benefits from a B-tree index

4. Recommend composite indexes for multi-column queries. Follow the equality-first, range-second ordering:

   • Place columns used with operators first in the index