Ikiru Yoshizaki

You can lint and fix GitHub Actions files with seiton, enforce .NET best practices and allocation-conscious performance rules, run benchmark-driven micro-optimization with disassembly analysis, conduct adversarial multi-agent code reviews, audit input surfaces for state-integrity issues, and govern specification documents under .github/docs/.

They target .NET/C# engineers optimizing hot paths, maintainers of GitHub Actions CI pipelines, and senior developers who need rigorous adversarial review, robustness audits of external input surfaces, and disciplined specification writing. Familiarity with C#, benchmarking, and CI configuration is assumed.

dotnet-performance-discipline supplies standing rules for ArrayPool ownership, Span/Memory selection, stackalloc budgets, and hot-path prohibitions. dotnet-micro-optimization then runs the measurement loop: one-hypothesis-per-variant benchmarks, JIT/native disassembly reading, and iteration until gains fall below 2-3%.

code-review examines implementation changes for correctness, performance, API usability, and test coverage, refuting its own findings before reporting. adversarial-audit instead exhaustively maps external state-restoration entry points and converts surviving invariants into unit tests, targeting systemic robustness rather than single changes.

Yes. The seiton skill requires the seiton CLI for GitHub Actions linting. The .NET skills assume a C# solution with benchmark harnesses such as BenchmarkDotNet. document-spec-policy expects a repository using the .github/docs/ specification structure, and grill-me disables model invocation by design.