القائمة
Use this skill when presenting benchmark results that compare the generated skillgraph nanousd backend against nanousd's default backend or another baseline backend.
التثبيت باستخدام Codex أو Claude انسخ هذا Prompt والصقه في Codex أو Claude أو مساعد آخر ليراجع صفحة Skill ويثبّتها لك.
استنادا إلى تصنيف SOC المهني
| name | nanousd-benchmark-reporting |
| description | Use this skill when presenting benchmark results that compare the generated skillgraph nanousd backend against nanousd's default backend or another baseline backend. |
| metadata | {"author":"NVIDIA"} |
Use this skill when presenting benchmark results that compare the generated skillgraph nanousd backend against nanousd's default backend or another baseline backend.
harness/nanousd_backend_benchmark.pyThis skill owns:
contracts/capabilities/nanousd-benchmark-reporting.jsonharness/nanousd_backend_report.pydocs/nanousd-backend-benchmarks.mdThe reporter consumes JSON emitted by harness/nanousd_backend_benchmark.py.
It must not rerun benchmarks, parse benchmark stdout, or silently change the
selection metric chosen by the benchmark runner. The benchmark runner owns
measurement and best-run selection; the report owns presentation.
Reports should include:
Use Markdown as the durable review artifact. SVG charts are useful for quick visual comparison, but they should supplement tables rather than replace them.
The default chart metric is open_ms, because open cost is currently the
highest-pressure signal for the generated backend. Use
rss_delta_bytes_per_prim or traverse_ns_per_prim when reviewing memory or
traversal-specific work.
Optional SLOW comparisons are performance pressure, not PR failure. Required
comparisons should be treated as gates only when a performance contract makes
that requirement explicit.
Do not track generated report outputs under reports/; they are local review
artifacts.
Do not conflate nanousd default backend attribute counts with generated authored attribute counts unless a correctness contract says they should match. Report the observed values and let the relevant backend contract decide whether a difference is meaningful.
Do not hide missing or failed backend results. A report may still render partial data, but it must make missing backend comparisons visible.
harness/nanousd_backend_benchmark.py --json-outMap unfamiliar or multi-repository codebases before refactoring, integration, or feature work. Use when Codex needs to analyze submodules, ownership boundaries, APIs, duplicate implementations, build/test surfaces, risk areas, and a practical refactor sequence before changing code.
Coordinate multiple AI agents or subagents on software work without duplicating effort or causing conflicts. Use when Codex is explicitly asked to delegate exploration, implementation, review, testing, integration, or parallel analysis across a codebase.
Maintain C ABIs and language bindings as exact, testable contracts. Use when evolving public headers, ctypes/cffi/pybind/Rust/JS bindings, shared libraries, struct layouts, GPU/native APIs, or any cross-language boundary where drift can cause crashes or silent corruption.
Debug rendering, viewer, shader, material, camera, lighting, and GPU integration bugs with reproducible visual evidence. Use when scenes render blank, only appear after camera motion, show incorrect shading, diverge across backends, crash in GUI mode, or need pixel/screenshot verification.
Refactor disjoint or duplicated projects into clear layered architecture with canonical contracts and thin adapters. Use when viewers, renderers, bindings, data libraries, plugins, or services overlap and need maintainable ownership boundaries without losing behavior.
Identify and remove obsolete legacy, fallback, staging, compatibility, duplicate, generated, and reference code before creating clean initial commits or new project history. Use when a repo is being reset, consolidated, or prepared for a cleaner public history.