bench-check

You can detect performance regressions by running benchmark suites that compare current metrics against a saved baseline. Bench-check executes build time, incremental rebuild, query depth, and embedding recall benchmarks to identify slowdowns beyond a set threshold.

Benchmark baseline comparison is the process of saving initial performance metrics and evaluating subsequent test runs against them. You need it to prevent silent slowdowns across releases by detecting regressions in build time or query performance.

You can update the benchmark baseline by running the bench-check tool to save a new baseline or overwrite the existing one. This replaces the saved metrics at generated/bench-check/baseline.json with your current optimized performance results.

Yes, you need a Node.js environment to run the benchmark scripts. Bench-check relies on scripts like benchmark.js, incremental-benchmark.js, query-benchmark.js, and embedding-benchmark.js to execute tests and generate JSON metrics with human-readable reports.

A performance regression report includes metrics for build time, incremental rebuilds, query depth, and embedding recall. The tool outputs a JSON metrics object alongside a human-readable report detailing any detected regressions beyond the threshold.

No, bench-check requires a saved baseline at generated/bench-check/baseline.json to compare current benchmark results. Without an existing baseline, you must first run the tool to save initial metrics before detecting future regressions.