perfgate (CLI crate)

Command-line entrypoint for perfgate.

This crate wires Clap commands to application use-cases in perfgate-app, handles JSON/file I/O, and enforces exit code policy for CI usage.

Commands

• run: execute a benchmark command and emit perfgate.run.v1.
• compare: compare current vs baseline and emit perfgate.compare.v1.
• md: render markdown from a compare receipt.
• github-annotations: emit GitHub Actions annotation lines.
• report: generate perfgate.report.v1 (optionally markdown too).
• promote: copy/normalize a run receipt into baseline storage.
• export: export run/compare data (csv, jsonl, html, prometheus).
• check: config-driven workflow for artifacts and gating.
• paired: interleaved baseline/current benchmarking for noise reduction.
• baseline: manage baselines on a centralized baseline server.

Quick Usage

perfgate run --name my-bench --out run.json -- ./my-benchmark

perfgate compare --baseline baseline.json --current run.json --out compare.json

perfgate md --compare compare.json --out comment.md

perfgate check --config perfgate.toml --bench my-bench

Exit Codes

• 0: success (or warn without --fail-on-warn)
• 1: tool/runtime error
• 2: policy fail
• 3: warn treated as failure (--fail-on-warn)

Scope

• This crate owns CLI UX, argument validation, and artifact file handling.
• It does not implement core policy math (domain) or process primitives (adapters).

More Documentation

• Workspace overview and CI examples: README.md
• Testing strategy: TESTING.md

License

Licensed under either Apache-2.0 or MIT.