`homeboy report`

Render reports from Homeboy structured output artifacts.

Synopsis

sh
homeboy report <COMMAND>

Subcommands

  • failure-digest — render a markdown failure digest from Homeboy command output JSON files
  • performance-digest — render resource, budget, baseline-health, preview, and Lab offload details from Homeboy run artifacts
  • bench-coverage — report list-only benchmark coverage for hot command paths
  • browser-evidence-compare — compare baseline/candidate browser evidence artifact bundles, optionally including visual screenshot diffs through a declared provider

Performance Digest

sh
homeboy report performance-digest 
  --output-dir <DIR> 
  [--metadata-json <JSON_OR_FILE>] 
  [--run-url <URL>] 
  [--min-samples 3] 
  [--max-cv-pct 20] 
  [--format markdown]

performance-digest reads run artifacts such as resource-summary.json and bench.json, then renders a Markdown summary of host pressure, extension child process usage, budget findings, benchmark memory peaks, baseline health, preview metadata, browser-origin evidence, and missing-artifact gaps. It is a report renderer only; it does not run benchmarks.

Browser Evidence Compare

sh
homeboy report browser-evidence-compare 
  --baseline-dir <DIR> 
  --candidate-dir <DIR> 
  --visual-compare 
  --visual-compare-provider <COMMAND> 
  --visual-artifacts-dir .homeboy/browser-visual-compare

--visual-compare pairs screenshot artifacts from matching scenario/profile/matrix variants and delegates diff generation to the executable named by --visual-compare-provider. The provider receives one input JSON path as its final argument and emits normalized visual compare JSON on stdout. Homeboy records mismatch metrics and artifact refs in the JSON report and Markdown output without knowing which extension or tool produced the diff.

Bench Coverage

sh
homeboy report bench-coverage [component] [--path <checkout>] [--all] [--format markdown|json]

bench-coverage uses the existing bench list/HOMEBOY_BENCH_LIST_ONLY=1 contract, so it discovers scenarios without running benchmarks. The report maps discovered scenarios onto generic hot command families such as audit, bench, lint, test, trace, refactor, runner, and offload, then shows which paths are covered or missing per component.