Benchmark Radar on repositories your team actually ships.

Scan speed claims are only useful when the method is repeatable. Radar exposes cold, warm, full, and diff-scope commands so teams can measure their real workflow.

Install Radar Benchmark docs

Methodology first.

A benchmark should state what was scanned, how cache was handled, and whether the command represented local work or CI enforcement.

Cold run

Run with `--no-cache` to measure first-pass inventory, parsing, rules, and report preparation.

Warm run

Run the same command again to measure cache-assisted local feedback.

Diff run

Run with `--diff` or `--since main` to measure pull-request scope.

Record context

Capture OS, CPU, repository size, selected files, scan mode, and Radar version.

Commands to run

Use these modes on the same repository and publish the selected file count next to wall-clock duration.

Mode Command Scope Best for

Local fast scan radar scan . --quick local loop use for pre-commit feedback

Cold fast scan radar scan . --quick --no-cache cold baseline measure first-run cost

Full security scan radar scan . pre-PR security, SCA, code-health, slop

Diff gate radar scan . --diff --fail-on high PR scope quality gate before merge

Trend output makes runs comparable.

After local scans, Radar stores compact history records so you can compare duration and scores without preserving full reports.

Local measurement loop

Run cold and warm scans, then inspect local history.

radar scan . --quick --no-cache
radar scan . --quick
radar scan . --diff
radar trend --limit 10