Missing index candidates
Index audit
Postgres index health check for safer review
Database Optimization Tool reviews PostgreSQL index health from evidence-focused statistics, then separates evidence-backed index candidates from AI-assisted advisory notes.
Unused index candidates
Scan-heavy table evidence
Human-owned index decisions
Missing index review
The health check looks for scan-heavy tables, repeated query cost, high read pressure, and low index usage before treating a missing index as a review candidate.
Unused index review
Low idx_scan, index size, uniqueness, primary-key role, and statistics window coverage are considered before an index is flagged for human review. Blind drops are not recommended.
Evidence Gate for index advice
AI-assisted index ideas can be useful, but they do not become official findings unless PostgreSQL metrics support the claim and the report asks for EXPLAIN validation.
evidence-focused operating boundary
Database Optimization Tool does not create indexes, drop indexes, change planner settings, or tune PostgreSQL automatically. It provides advisory review material only.
FAQ
Frequently asked questions
These answers describe the product focus: careful database evidence, clear findings, and team-approved next steps.
Does the index health check create or drop indexes?
No. It reads PostgreSQL evidence and reports candidates. Creating or dropping indexes requires human review, EXPLAIN validation, and a production rollout plan.
Can an unused index finding be wrong?
Yes. A short statistics window can miss monthly jobs, rare admin workflows, or failover behavior. The report treats unused indexes as review candidates, not automatic drop instructions.
How are missing index candidates supported?
They need workload evidence such as repeated query cost, sequential scan pressure, read pressure, or EXPLAIN-ready query patterns. Unsupported suggestions remain advisory.
Is this a real-time index monitor?
No. It is a evidence-focused PostgreSQL audit workflow based on collected evidence, not continuous monitoring or alerting.