Live Trino SQL
Run real SQL against TPC-H sf=0.01 on a JVM Trino single-node server co-deployed on Oracle Cloud. Sample queries or your own — results and wall-clock latency.
Differential testing with Trino
Run any query through both engines side-by-side. Cell-level diff. We've deliberately seeded one query with a bug; the Trino comparison catches it, just like SQLancer does for real.
Agent pipeline simulator
Animated walk-through of one SPEC → IMPLEMENT → VERIFY → FUZZ → SHIP run with realistic artifacts at every stage boundary. Click any stage to drill in.
TPC-H benchmarks
Full TPC-H Q1–Q22 vs. Trino at scale factors 1, 5, and 10. Geomean speedups: 6.342× (SF1), 3.644× (SF5), 3.037× (SF10).
System design diagrams
Standalone Rust engine: Trino HTTP surface, native vectorized execution, Parquet/Iceberg catalog, and optional distributed coordinator/worker runtime.
Quick endpoints reference
Trino-compatible
POST /v1/statement
GET /v1/statement/{id}/{token}
DEL /v1/statement/{id}/{token}
Any Trino client — trino-cli, JDBC, the python trino package — can talk to this. Try it:
curl -s localhost:8787/v1/statement \ -d 'SELECT count(*) FROM lineitem' | jq
Demo helpers
GET /api/catalog · tables & schemas
GET /api/samples · seeded queries
POST /api/parse · parse + transpile
POST /api/explain · logical plan tree
POST /api/diff · differential vs Trino
POST /api/pipeline/run · agent pipeline run
GET /api/benchmarks · live timings
GET /version · release version & date