name: harness-benchmark description: "Creates benchmark scripts and threshold checks. Use when establishing performance baselines or regression gates. Do NOT use for ad-hoc Python profiling (use python-performance-optimization)." metadata: version: "1.0.1"
Harness Benchmark
Benchmarks must match the product's critical paths.
Discovery
- Identify core user or system flows from docs, routes, commands, or tests.
- Find existing benchmark, load test, Lighthouse, or timing scripts.
- Pick one critical path first.
- Define explicit pass/fail thresholds.
Possible Outputs
- benchmark runner script (
benchmark.ps1 or benchmark.sh) - package script proposal
- docs-only benchmark plan
Output
markdown
## Benchmark Proposal
Critical path: ...
Metric: ...
Threshold: ...
Command: ...
Data setup: ...
Approval needed before writing: yes
Rules
- Do not create benchmark scripts without approval.
- Thresholds must be explicit.
- Prefer repeatable local checks before external services.
- Record environment assumptions.