ARGUS CLI

SQL Stress-Test Auditor. Your SQL is fast — and we have proof.

ARGUS takes your schema and a query, generates realistic data at scale, benchmarks candidate optimizations against each other on your own machine, and reports which one actually wins — with hard numbers.

ARGUS currently supports PostgreSQL only.

Install

npm install -g @argusaudit/cli

Requirements

• Node.js >= 24
• Docker and Docker Compose >= 2.20 — benchmarks run in local containers. Compose 2.20 is the floor because the CLI drives the stack with docker compose wait.

Usage

argus login                                   # authenticate via browser
argus audit --ddl schema.sql --query slow.sql

The audit prints a link to your report when it finishes.

argus audit

| Flag | Purpose | | --- | --- | | --ddl <path> | Path to your DDL/schema file. Required. | | --query <path> | Path to the query to optimize. Required. | | --postgres-version <version> | Benchmark against a specific Postgres, e.g. 17, 17.5, 17-bookworm. Defaults to 18-alpine. | | --keep-containers | Leave containers running after the benchmark instead of tearing them down. Useful for debugging. |

Other commands

• argus login — authenticate via browser; credentials are stored in your OS keychain.
• argus logout — remove stored credentials.

How it works

1. Your DDL and query are sent to the ARGUS engine, which proposes optimization variants (index / query rewrite / schema change).
2. The CLI starts an isolated Postgres via Docker Compose and pulls the argusaudit/worker image.
3. The worker seeds data matching realistic distributions, then benchmarks each variant 5x with ROLLBACK between runs.
4. Results are reported back and rendered at argusaudit.dev.

Your data never leaves your machine — only the DDL and query go to the cloud. Benchmarking happens entirely in local containers.

License

MIT — see LICENSE.