Input

Create an isolated PostgreSQL 17 database with 200,000 synthetic orders and 10,000 customer IDs. Filter customer_id = 123. Compare EXPLAIN (ANALYZE, BUFFERS, FORMAT JSON) before and after CREATE INDEX orders_customer_id_idx ON orders (customer_id). Take one warmup and five measured runs per condition. Do not connect to a production database.

Acceptance criteria

Both result sets contain the same 20 rows. The baseline uses Seq Scan; the indexed plan uses orders_customer_id_idx. Record actual timings and limitations; no speedup threshold is required. Scope: one SQL recipe, not an agent benchmark.

Recorded outcome

Scope: direct SQL recipe validation, not an AI-agent end-to-end evaluation.
200,000 synthetic orders; customer_id = 123 matches 20 rows.
Before: Seq Scan. After: Bitmap Heap Scan with orders_customer_id_idx.
Result rows and SHA-256 match before and after.
Five warmed measurements: median 12.676 ms before; 0.152 ms after.
Timing is specific to this workload, cache, and hardware. Runs are sequential, not randomized.
No production queries, third-party scripts, or paid AI calls were used.
Only the missing-index recipe was checked; no full-skill effectiveness or safety certification is implied.
Result SHA-256: f72cd8761469a9f1c53d4d7fd478d05857f685a5395e7300ae98ad696cdac8fb
Probe script SHA-256: 0a429cdfed4779f645767cd945d08cda58df801be271037ae362cb2f191337b7

Environment

PostgreSQL 17.11 on x86_64-pc-linux-musl, compiled by gcc (Alpine 15.2.0) 15.2.0, 64-bit; Docker isolated database; 1 CPU and 512 MiB database limit; parallel query disabled.

Package SHA-256: 2e93a64c76cbde8d4dd1b7d8c6896f21492d0425b8ff25fcc69d97769b4f0e3d

Outcome recorded: 2026-10-02 17:17 UTC