offload-benchmark

You can automate benchmark timing and README updates by running a skill that compares local and Offload performance, collects wall-clock time and speedup metrics, and replaces the README Benchmarks section with the generated data.

To compare local and Offload pytest performance for mng and sculptor, run a benchmarking skill that collects precise timing data, computes speedups, and identifies test counts from the pytest outputs.

Yes, you can generate and insert benchmark metrics into your project README by computing time, percentage, and speedup data, then feeding those metrics into a template that replaces the Offload README Benchmarks section.

Invalid benchmark runs are flagged during automated pytest timing when tests fail, ensuring the generated README notes the INVALID results while preserving the rest of the document.

No external dependencies are required to run mng and sculptor benchmarks, as the skill operates without any listed dependencies to compare local and Offload performance and update the README.