Sample report · no account needed

This is what you get

A real report, generated by the same code that analyses customer databases, from a synthetic shop database with six slow statements. Expand a card to see its plan, table and index statistics and the visual plan; every button works.

Open full screen Start an analysis

What to look at first

Cards are ranked by total time: calls × mean. The first card, an order lookup, has a mean of 412 ms and runs 48,000 times an hour; that is the statement to fix on Monday.

How to read a card

The metric grid shows calls, mean, min, max, cache hit ratio and blocks read. Below it: the statement, the plan, then the tables it touches with their size, dead-tuple ratio and every index with its scan count.

Visual plan and AI prompts

“Visual Plan” opens the plan in PEV2 with the slowest node highlighted. “Ask Claude” and “Ask Copilot” copy a prompt with the plan and statistics already filled in.

The same report for your database, in about a minute

  1. 0:00Paste a connection string Any PostgreSQL: Supabase, Neon, RDS, your own server. The password is used once and never stored.
  2. 0:10Pick the thresholds Minimum calls and minimum mean time decide what counts as slow. Defaults work for most databases.
  3. 0:20The run starts Read-only queries rank the statements from pg_stat_statements and collect execution plans and table statistics.
  4. 0:35Open the report One HTML file: cards ranked by total time, each with its plan, table and index statistics and a PEV2 visual plan.
  5. 0:45Read the recommendations Index suggestions with estimated benefit, rewrite hints and what changed since the last run.
  6. 0:55Share or schedule Send an expiring link to a colleague, export a PDF, or schedule the analysis weekly and get told about regressions.

Read-only throughout; the password is used once and never stored. What we usually find · Security & trust.